id sid tid token lemma pos bracis-19054 1 1 a a DET bracis-19054 1 2 heterogeneous heterogeneous ADJ bracis-19054 1 3 network network NOUN bracis-19054 1 4 - - PUNCT bracis-19054 1 5 based base VERB bracis-19054 1 6 positive positive ADJ bracis-19054 1 7 and and CCONJ bracis-19054 1 8 unlabeled unlabeled ADJ bracis-19054 1 9 learning learn VERB bracis-19054 1 10 approach approach NOUN bracis-19054 1 11 to to PART bracis-19054 1 12 detect detect VERB bracis-19054 1 13 fake fake ADJ bracis-19054 1 14 news news NOUN bracis-19054 1 15 | | NOUN bracis-19054 1 16 springer springer NOUN bracis-19054 1 17 nature nature PROPN bracis-19054 1 18 link link PROPN bracis-19054 1 19 ( ( PUNCT bracis-19054 1 20 formerly formerly ADV bracis-19054 1 21 springerlink springerlink NOUN bracis-19054 1 22 ) ) PUNCT bracis-19054 1 23 skip skip VERB bracis-19054 1 24 to to ADP bracis-19054 1 25 main main ADJ bracis-19054 1 26 content content NOUN bracis-19054 1 27 advertisement advertisement NOUN bracis-19054 1 28 log log NOUN bracis-19054 1 29 in in ADP bracis-19054 1 30 menu menu NOUN bracis-19054 1 31 find find VERB bracis-19054 1 32 a a DET bracis-19054 1 33 journal journal NOUN bracis-19054 1 34 publish publish VERB bracis-19054 1 35 with with ADP bracis-19054 1 36 us we PRON bracis-19054 1 37 track track VERB bracis-19054 1 38 your your PRON bracis-19054 1 39 research research NOUN bracis-19054 1 40 search search NOUN bracis-19054 1 41 cart cart NOUN bracis-19054 1 42 home home NOUN bracis-19054 1 43 intelligent intelligent ADJ bracis-19054 1 44 systems system NOUN bracis-19054 1 45 conference conference NOUN bracis-19054 1 46 paper paper NOUN bracis-19054 1 47 a a DET bracis-19054 1 48 heterogeneous heterogeneous ADJ bracis-19054 1 49 network network NOUN bracis-19054 1 50 - - PUNCT bracis-19054 1 51 based base VERB bracis-19054 1 52 positive positive ADJ bracis-19054 1 53 and and CCONJ bracis-19054 1 54 unlabeled unlabeled ADJ bracis-19054 1 55 learning learn VERB bracis-19054 1 56 approach approach NOUN bracis-19054 1 57 to to PART bracis-19054 1 58 detect detect VERB bracis-19054 1 59 fake fake ADJ bracis-19054 1 60 news news NOUN bracis-19054 1 61 conference conference NOUN bracis-19054 1 62 paper paper NOUN bracis-19054 1 63 first first ADV bracis-19054 1 64 online online ADV bracis-19054 1 65 : : PUNCT bracis-19054 1 66 28 28 NUM bracis-19054 1 67 november november NOUN bracis-19054 1 68 2021 2021 NUM bracis-19054 1 69 pp pp ADP bracis-19054 2 1 3–18 3–18 PROPN bracis-19054 2 2 cite cite VERB bracis-19054 2 3 this this DET bracis-19054 2 4 conference conference NOUN bracis-19054 2 5 paper paper NOUN bracis-19054 2 6 access access NOUN bracis-19054 2 7 provided provide VERB bracis-19054 2 8 by by ADP bracis-19054 2 9 university university PROPN bracis-19054 2 10 of of ADP bracis-19054 2 11 notre notre PROPN bracis-19054 2 12 dame dame PROPN bracis-19054 2 13 hesburgh hesburgh PROPN bracis-19054 2 14 library library PROPN bracis-19054 2 15 download download PROPN bracis-19054 2 16 book book NOUN bracis-19054 2 17 pdf pdf PROPN bracis-19054 2 18 download download NOUN bracis-19054 2 19 book book NOUN bracis-19054 2 20 epub epub PROPN bracis-19054 2 21 intelligent intelligent ADJ bracis-19054 2 22 systems system NOUN bracis-19054 2 23 ( ( PUNCT bracis-19054 2 24 bracis bracis NOUN bracis-19054 2 25 2021 2021 NUM bracis-19054 2 26 ) ) PUNCT bracis-19054 2 27 a a DET bracis-19054 2 28 heterogeneous heterogeneous ADJ bracis-19054 2 29 network network NOUN bracis-19054 2 30 - - PUNCT bracis-19054 2 31 based base VERB bracis-19054 2 32 positive positive ADJ bracis-19054 2 33 and and CCONJ bracis-19054 2 34 unlabeled unlabeled ADJ bracis-19054 2 35 learning learn VERB bracis-19054 2 36 approach approach NOUN bracis-19054 2 37 to to PART bracis-19054 2 38 detect detect VERB bracis-19054 2 39 fake fake ADJ bracis-19054 2 40 news news NOUN bracis-19054 2 41 download download NOUN bracis-19054 2 42 book book NOUN bracis-19054 2 43 pdf pdf PROPN bracis-19054 2 44 download download NOUN bracis-19054 2 45 book book PROPN bracis-19054 2 46 epub epub PROPN bracis-19054 2 47 mariana mariana PROPN bracis-19054 2 48 c. c. PROPN bracis-19054 2 49 de de PROPN bracis-19054 2 50 souza souza PROPN bracis-19054 2 51     SPACE bracis-19054 2 52 orcid orcid PROPN bracis-19054 2 53 : : PUNCT bracis-19054 2 54 orcid.org/0000-0002-1746-841410 orcid.org/0000-0002-1746-841410 PROPN bracis-19054 2 55 , , PUNCT bracis-19054 2 56 bruno bruno PROPN bracis-19054 2 57 m. m. PROPN bracis-19054 2 58 nogueira nogueira PROPN bracis-19054 2 59     SPACE bracis-19054 2 60 orcid orcid PROPN bracis-19054 2 61 : : PUNCT bracis-19054 2 62 orcid.org/0000-0002-2401-242311 orcid.org/0000-0002-2401-242311 NOUN bracis-19054 2 63 , , PUNCT bracis-19054 2 64 rafael rafael PROPN bracis-19054 2 65 g. g. PROPN bracis-19054 2 66 rossi rossi PROPN bracis-19054 2 67     SPACE bracis-19054 2 68 orcid orcid NOUN bracis-19054 2 69 : : PUNCT bracis-19054 2 70 orcid.org/0000-0001-8513-321312 orcid.org/0000-0001-8513-321312 NOUN bracis-19054 2 71 , , PUNCT bracis-19054 2 72 ricardo ricardo PROPN bracis-19054 2 73 m. m. NOUN bracis-19054 2 74 marcacini marcacini NOUN bracis-19054 2 75     SPACE bracis-19054 2 76 orcid orcid NOUN bracis-19054 2 77 : : PUNCT bracis-19054 2 78 orcid.org/0000-0002-2309-348710 orcid.org/0000-0002-2309-348710 PROPN bracis-19054 2 79 & & CCONJ bracis-19054 2 80 … … PUNCT bracis-19054 2 81 solange solange PROPN bracis-19054 2 82 o. o. PROPN bracis-19054 2 83 rezende rezende PROPN bracis-19054 2 84     SPACE bracis-19054 2 85 orcid orcid NOUN bracis-19054 2 86 : : PUNCT bracis-19054 2 87 orcid.org/0000-0002-5233-763910 orcid.org/0000-0002-5233-763910 NOUN bracis-19054 2 88     SPACE bracis-19054 2 89 show show VERB bracis-19054 2 90 authors author NOUN bracis-19054 2 91 part part NOUN bracis-19054 2 92 of of ADP bracis-19054 2 93 the the DET bracis-19054 2 94 book book NOUN bracis-19054 2 95 series series NOUN bracis-19054 2 96 : : PUNCT bracis-19054 2 97 lecture lecture NOUN bracis-19054 2 98 notes note NOUN bracis-19054 2 99 in in ADP bracis-19054 2 100 computer computer NOUN bracis-19054 2 101 science science NOUN bracis-19054 2 102 ( ( PUNCT bracis-19054 2 103 ( ( PUNCT bracis-19054 2 104 lnai lnai ADJ bracis-19054 2 105 , , PUNCT bracis-19054 2 106 volume volume NOUN bracis-19054 2 107 13074 13074 NUM bracis-19054 2 108 ) ) PUNCT bracis-19054 2 109 ) ) PUNCT bracis-19054 2 110 included include VERB bracis-19054 2 111 in in ADP bracis-19054 2 112 the the DET bracis-19054 2 113 following follow VERB bracis-19054 2 114 conference conference NOUN bracis-19054 2 115 series series NOUN bracis-19054 2 116 : : PUNCT bracis-19054 2 117 brazilian brazilian ADJ bracis-19054 2 118 conference conference NOUN bracis-19054 2 119 on on ADP bracis-19054 2 120 intelligent intelligent ADJ bracis-19054 2 121 systems system NOUN bracis-19054 2 122 1271 1271 NUM bracis-19054 2 123 accesses access VERB bracis-19054 2 124 8 8 NUM bracis-19054 2 125 citations citation NOUN bracis-19054 2 126 abstract abstract VERB bracis-19054 2 127 the the DET bracis-19054 2 128 dynamism dynamism NOUN bracis-19054 2 129 of of ADP bracis-19054 2 130 fake fake ADJ bracis-19054 2 131 news news NOUN bracis-19054 2 132 evolution evolution NOUN bracis-19054 2 133 and and CCONJ bracis-19054 2 134 dissemination dissemination NOUN bracis-19054 2 135 plays play VERB bracis-19054 2 136 a a DET bracis-19054 2 137 crucial crucial ADJ bracis-19054 2 138 role role NOUN bracis-19054 2 139 in in ADP bracis-19054 2 140 influencing influence VERB bracis-19054 2 141 and and CCONJ bracis-19054 2 142 confirming confirm VERB bracis-19054 2 143 personal personal ADJ bracis-19054 2 144 beliefs belief NOUN bracis-19054 2 145 . . PUNCT bracis-19054 3 1 to to PART bracis-19054 3 2 minimize minimize VERB bracis-19054 3 3 the the DET bracis-19054 3 4 spread spread NOUN bracis-19054 3 5 of of ADP bracis-19054 3 6 disinformation disinformation NOUN bracis-19054 3 7 approaches approach NOUN bracis-19054 3 8 proposed propose VERB bracis-19054 3 9 in in ADP bracis-19054 3 10 the the DET bracis-19054 3 11 literature literature NOUN bracis-19054 3 12 , , PUNCT bracis-19054 3 13 automatic automatic ADJ bracis-19054 3 14 fake fake ADJ bracis-19054 3 15 news news NOUN bracis-19054 3 16 detection detection NOUN bracis-19054 3 17 generally generally ADV bracis-19054 3 18 learns learn VERB bracis-19054 3 19 models model NOUN bracis-19054 3 20 through through ADP bracis-19054 3 21 binary binary ADJ bracis-19054 3 22 supervised supervised ADJ bracis-19054 3 23 algorithms algorithm NOUN bracis-19054 3 24 considering consider VERB bracis-19054 3 25 textual textual ADJ bracis-19054 3 26 and and CCONJ bracis-19054 3 27 contextual contextual ADJ bracis-19054 3 28 information information NOUN bracis-19054 3 29 . . PUNCT bracis-19054 4 1 however however ADV bracis-19054 4 2 , , PUNCT bracis-19054 4 3 labeling label VERB bracis-19054 4 4 significant significant ADJ bracis-19054 4 5 amounts amount NOUN bracis-19054 4 6 of of ADP bracis-19054 4 7 real real ADJ bracis-19054 4 8 news news NOUN bracis-19054 4 9 to to PART bracis-19054 4 10 build build VERB bracis-19054 4 11 accurate accurate ADJ bracis-19054 4 12 classifiers classifier NOUN bracis-19054 4 13 is be AUX bracis-19054 4 14 difficult difficult ADJ bracis-19054 4 15 and and CCONJ bracis-19054 4 16 time time NOUN bracis-19054 4 17 - - PUNCT bracis-19054 4 18 consuming consume VERB bracis-19054 4 19 due due ADP bracis-19054 4 20 to to ADP bracis-19054 4 21 their their PRON bracis-19054 4 22 broad broad ADJ bracis-19054 4 23 spectrum spectrum NOUN bracis-19054 4 24 . . PUNCT bracis-19054 5 1 positive positive ADJ bracis-19054 5 2 and and CCONJ bracis-19054 5 3 unlabeled unlabeled ADJ bracis-19054 5 4 learning learning NOUN bracis-19054 5 5 ( ( PUNCT bracis-19054 5 6 pul pul NOUN bracis-19054 5 7 ) ) PUNCT bracis-19054 5 8 can can AUX bracis-19054 5 9 be be AUX bracis-19054 5 10 a a DET bracis-19054 5 11 good good ADJ bracis-19054 5 12 alternative alternative NOUN bracis-19054 5 13 in in ADP bracis-19054 5 14 this this DET bracis-19054 5 15 scenario scenario NOUN bracis-19054 5 16 . . PUNCT bracis-19054 6 1 pul pul NOUN bracis-19054 6 2 algorithms algorithm NOUN bracis-19054 6 3 learn learn VERB bracis-19054 6 4 models model NOUN bracis-19054 6 5 considering consider VERB bracis-19054 6 6 little little ADJ bracis-19054 6 7 labeled label VERB bracis-19054 6 8 data datum NOUN bracis-19054 6 9 of of ADP bracis-19054 6 10 the the DET bracis-19054 6 11 interest interest NOUN bracis-19054 6 12 class class NOUN bracis-19054 6 13 and and CCONJ bracis-19054 6 14 use use VERB bracis-19054 6 15 unlabeled unlabeled ADJ bracis-19054 6 16 data datum NOUN bracis-19054 6 17 to to PART bracis-19054 6 18 increase increase VERB bracis-19054 6 19 classification classification NOUN bracis-19054 6 20 performance performance NOUN bracis-19054 6 21 . . PUNCT bracis-19054 7 1 this this DET bracis-19054 7 2 paper paper NOUN bracis-19054 7 3 proposes propose VERB bracis-19054 7 4 a a DET bracis-19054 7 5 heterogeneous heterogeneous ADJ bracis-19054 7 6 network network NOUN bracis-19054 7 7 variant variant NOUN bracis-19054 7 8 of of ADP bracis-19054 7 9 the the DET bracis-19054 7 10 pu pu PROPN bracis-19054 7 11 - - PUNCT bracis-19054 7 12 lp lp PROPN bracis-19054 7 13 algorithm algorithm NOUN bracis-19054 7 14 , , PUNCT bracis-19054 7 15 a a DET bracis-19054 7 16 pul pul NOUN bracis-19054 7 17 algorithm algorithm NOUN bracis-19054 7 18 based base VERB bracis-19054 7 19 on on ADP bracis-19054 7 20 similarity similarity NOUN bracis-19054 7 21 networks network NOUN bracis-19054 7 22 . . PUNCT bracis-19054 8 1 our our PRON bracis-19054 8 2 network network NOUN bracis-19054 8 3 incorporates incorporate VERB bracis-19054 8 4 different different ADJ bracis-19054 8 5 linguistic linguistic ADJ bracis-19054 8 6 features feature NOUN bracis-19054 8 7 to to PART bracis-19054 8 8 characterize characterize VERB bracis-19054 8 9 fake fake ADJ bracis-19054 8 10 news news NOUN bracis-19054 8 11 , , PUNCT bracis-19054 8 12 such such ADJ bracis-19054 8 13 as as ADP bracis-19054 8 14 representative representative ADJ bracis-19054 8 15 terms term NOUN bracis-19054 8 16 , , PUNCT bracis-19054 8 17 emotiveness emotiveness NOUN bracis-19054 8 18 , , PUNCT bracis-19054 8 19 pausality pausality NOUN bracis-19054 8 20 , , PUNCT bracis-19054 8 21 and and CCONJ bracis-19054 8 22 average average ADJ bracis-19054 8 23 sentence sentence NOUN bracis-19054 8 24 size size NOUN bracis-19054 8 25 . . PUNCT bracis-19054 9 1 also also ADV bracis-19054 9 2 , , PUNCT bracis-19054 9 3 we we PRON bracis-19054 9 4 considered consider VERB bracis-19054 9 5 two two NUM bracis-19054 9 6 representations representation NOUN bracis-19054 9 7 of of ADP bracis-19054 9 8 the the DET bracis-19054 9 9 news news NOUN bracis-19054 9 10 to to PART bracis-19054 9 11 compute compute VERB bracis-19054 9 12 similarity similarity NOUN bracis-19054 9 13 : : PUNCT bracis-19054 9 14 term term NOUN bracis-19054 9 15 frequency frequency NOUN bracis-19054 9 16 - - PUNCT bracis-19054 9 17 inverse inverse NOUN bracis-19054 9 18 document document NOUN bracis-19054 9 19 frequency frequency NOUN bracis-19054 9 20 , , PUNCT bracis-19054 9 21 and and CCONJ bracis-19054 9 22 doc2vec doc2vec PROPN bracis-19054 9 23 , , PUNCT bracis-19054 9 24 which which PRON bracis-19054 9 25 creates create VERB bracis-19054 9 26 fixed fix VERB bracis-19054 9 27 - - PUNCT bracis-19054 9 28 sized sized ADJ bracis-19054 9 29 document document NOUN bracis-19054 9 30 representations representation NOUN bracis-19054 9 31 regardless regardless ADV bracis-19054 9 32 of of ADP bracis-19054 9 33 its its PRON bracis-19054 9 34 length length NOUN bracis-19054 9 35 . . PUNCT bracis-19054 10 1 we we PRON bracis-19054 10 2 evaluated evaluate VERB bracis-19054 10 3 our our PRON bracis-19054 10 4 approach approach NOUN bracis-19054 10 5 in in ADP bracis-19054 10 6 six six NUM bracis-19054 10 7 datasets dataset NOUN bracis-19054 10 8 written write VERB bracis-19054 10 9 in in ADP bracis-19054 10 10 portuguese portuguese PROPN bracis-19054 10 11 or or CCONJ bracis-19054 10 12 english english PROPN bracis-19054 10 13 , , PUNCT bracis-19054 10 14 comparing compare VERB bracis-19054 10 15 its its PRON bracis-19054 10 16 performance performance NOUN bracis-19054 10 17 with with ADP bracis-19054 10 18 a a DET bracis-19054 10 19 binary binary ADJ bracis-19054 10 20 semi semi ADV bracis-19054 10 21 - - ADJ bracis-19054 10 22 supervised supervised ADJ bracis-19054 10 23 baseline baseline NOUN bracis-19054 10 24 algorithm algorithm NOUN bracis-19054 10 25 , , PUNCT bracis-19054 10 26 using use VERB bracis-19054 10 27 two two NUM bracis-19054 10 28 well well ADV bracis-19054 10 29 - - PUNCT bracis-19054 10 30 established establish VERB bracis-19054 10 31 label label NOUN bracis-19054 10 32 propagation propagation NOUN bracis-19054 10 33 algorithms algorithm NOUN bracis-19054 10 34 : : PUNCT bracis-19054 10 35 lphn lphn PROPN bracis-19054 10 36 and and CCONJ bracis-19054 10 37 gnetmine gnetmine ADJ bracis-19054 10 38 . . PUNCT bracis-19054 11 1 the the DET bracis-19054 11 2 results result NOUN bracis-19054 11 3 indicate indicate VERB bracis-19054 11 4 that that SCONJ bracis-19054 11 5 pu pu PROPN bracis-19054 11 6 - - PUNCT bracis-19054 11 7 lp lp PROPN bracis-19054 11 8 with with ADP bracis-19054 11 9 heterogeneous heterogeneous ADJ bracis-19054 11 10 networks network NOUN bracis-19054 11 11 can can AUX bracis-19054 11 12 be be AUX bracis-19054 11 13 competitive competitive ADJ bracis-19054 11 14 to to ADP bracis-19054 11 15 binary binary PROPN bracis-19054 11 16 semi semi ADJ bracis-19054 11 17 - - ADJ bracis-19054 11 18 supervised supervised ADJ bracis-19054 11 19 learning learning NOUN bracis-19054 11 20 . . PUNCT bracis-19054 12 1 also also ADV bracis-19054 12 2 , , PUNCT bracis-19054 12 3 linguistic linguistic ADJ bracis-19054 12 4 features feature NOUN bracis-19054 12 5 such such ADJ bracis-19054 12 6 as as ADP bracis-19054 12 7 representative representative ADJ bracis-19054 12 8 terms term NOUN bracis-19054 12 9 and and CCONJ bracis-19054 12 10 pausality pausality NOUN bracis-19054 12 11 improved improve VERB bracis-19054 12 12 the the DET bracis-19054 12 13 classification classification NOUN bracis-19054 12 14 performance performance NOUN bracis-19054 12 15 , , PUNCT bracis-19054 12 16 especially especially ADV bracis-19054 12 17 when when SCONJ bracis-19054 12 18 there there PRON bracis-19054 12 19 is be VERB bracis-19054 12 20 a a DET bracis-19054 12 21 small small ADJ bracis-19054 12 22 amount amount NOUN bracis-19054 12 23 of of ADP bracis-19054 12 24 labeled label VERB bracis-19054 12 25 news news NOUN bracis-19054 12 26 . . PUNCT bracis-19054 13 1 supported support VERB bracis-19054 13 2 by by ADP bracis-19054 13 3 coordenação coordenação PROPN bracis-19054 13 4 de de PROPN bracis-19054 13 5 aperfeiçoamento aperfeiçoamento PROPN bracis-19054 13 6 de de X bracis-19054 13 7 pessoal pessoal PROPN bracis-19054 13 8 de de PROPN bracis-19054 13 9 nível nível PROPN bracis-19054 13 10 superior superior NOUN bracis-19054 14 1 [ [ X bracis-19054 14 2 10662147 10662147 NUM bracis-19054 14 3 / / SYM bracis-19054 14 4 d d NOUN bracis-19054 14 5 ] ] X bracis-19054 14 6 , , PUNCT bracis-19054 14 7 fundação fundação PROPN bracis-19054 14 8 de de X bracis-19054 14 9 amparo amparo PROPN bracis-19054 14 10 à à X bracis-19054 14 11 pesquisa pesquisa PROPN bracis-19054 14 12 do do AUX bracis-19054 14 13 estado estado PROPN bracis-19054 14 14 de de PROPN bracis-19054 14 15 são são PROPN bracis-19054 14 16 paulo paulo PROPN bracis-19054 15 1 [ [ X bracis-19054 15 2 2019/25010 2019/25010 NUM bracis-19054 15 3 - - SYM bracis-19054 15 4 5 5 NUM bracis-19054 15 5 , , PUNCT bracis-19054 15 6 2019/07665 2019/07665 NUM bracis-19054 15 7 - - SYM bracis-19054 15 8 4 4 NUM bracis-19054 15 9 ] ] PUNCT bracis-19054 15 10 , , PUNCT bracis-19054 15 11 and and CCONJ bracis-19054 15 12 conselho conselho VERB bracis-19054 15 13 nacional nacional PROPN bracis-19054 15 14 de de X bracis-19054 15 15 desenvolvimento desenvolvimento PROPN bracis-19054 15 16 científico científico PROPN bracis-19054 15 17 e e NOUN bracis-19054 15 18 tecnológico tecnológico PROPN bracis-19054 15 19 [ [ X bracis-19054 15 20 426663/2018 426663/2018 NUM bracis-19054 15 21 - - SYM bracis-19054 15 22 7 7 NUM bracis-19054 15 23 , , PUNCT bracis-19054 15 24 433082/2018 433082/2018 NUM bracis-19054 15 25 - - SYM bracis-19054 15 26 6 6 NUM bracis-19054 15 27 , , PUNCT bracis-19054 15 28 and and CCONJ bracis-19054 15 29 438017/2018 438017/2018 NUM bracis-19054 15 30 - - SYM bracis-19054 15 31 8 8 NUM bracis-19054 15 32 ] ] PUNCT bracis-19054 15 33 . . PUNCT bracis-19054 16 1 access access NOUN bracis-19054 16 2 provided provide VERB bracis-19054 16 3 by by ADP bracis-19054 16 4 university university PROPN bracis-19054 16 5 of of ADP bracis-19054 16 6 notre notre PROPN bracis-19054 16 7 dame dame PROPN bracis-19054 16 8 hesburgh hesburgh PROPN bracis-19054 16 9 library library PROPN bracis-19054 16 10 . . PUNCT bracis-19054 17 1 download download PROPN bracis-19054 17 2 conference conference NOUN bracis-19054 17 3 paper paper NOUN bracis-19054 17 4 pdf pdf NOUN bracis-19054 17 5 similar similar ADJ bracis-19054 17 6 content content NOUN bracis-19054 17 7 being be AUX bracis-19054 17 8 viewed view VERB bracis-19054 17 9 by by ADP bracis-19054 17 10 others other NOUN bracis-19054 17 11 a a DET bracis-19054 17 12 network network NOUN bracis-19054 17 13 - - PUNCT bracis-19054 17 14 based base VERB bracis-19054 17 15 positive positive ADJ bracis-19054 17 16 and and CCONJ bracis-19054 17 17 unlabeled unlabeled ADJ bracis-19054 17 18 learning learn VERB bracis-19054 17 19 approach approach NOUN bracis-19054 17 20 for for ADP bracis-19054 17 21 fake fake ADJ bracis-19054 17 22 news news NOUN bracis-19054 17 23 detection detection NOUN bracis-19054 17 24 article article NOUN bracis-19054 17 25 18 18 NUM bracis-19054 17 26 november november PROPN bracis-19054 17 27 2021 2021 NUM bracis-19054 17 28 fake fake ADJ bracis-19054 17 29 news news NOUN bracis-19054 17 30 detection detection NOUN bracis-19054 17 31 chapter chapter NOUN bracis-19054 17 32 © © PROPN bracis-19054 17 33 2021 2021 NUM bracis-19054 17 34 positive positive ADJ bracis-19054 17 35 - - PUNCT bracis-19054 17 36 unlabeled unlabeled ADJ bracis-19054 17 37 learning learning NOUN bracis-19054 17 38 for for ADP bracis-19054 17 39 sentiment sentiment NOUN bracis-19054 17 40 analysis analysis NOUN bracis-19054 17 41 with with ADP bracis-19054 17 42 adversarial adversarial ADJ bracis-19054 17 43 training training NOUN bracis-19054 17 44 chapter chapter NOUN bracis-19054 17 45 © © PROPN bracis-19054 17 46 2019 2019 NUM bracis-19054 17 47 explore explore VERB bracis-19054 17 48 related relate VERB bracis-19054 17 49 subjects subject NOUN bracis-19054 17 50 discover discover VERB bracis-19054 17 51 the the DET bracis-19054 17 52 latest late ADJ bracis-19054 17 53 articles article NOUN bracis-19054 17 54 , , PUNCT bracis-19054 17 55 books book NOUN bracis-19054 17 56 and and CCONJ bracis-19054 17 57 news news NOUN bracis-19054 17 58 in in ADP bracis-19054 17 59 related related ADJ bracis-19054 17 60 subjects subject NOUN bracis-19054 17 61 , , PUNCT bracis-19054 17 62 suggested suggest VERB bracis-19054 17 63 using use VERB bracis-19054 17 64 machine machine NOUN bracis-19054 17 65 learning learning NOUN bracis-19054 17 66 . . PUNCT bracis-19054 18 1 categorization categorization NOUN bracis-19054 18 2 heterodox heterodox NOUN bracis-19054 18 3 economics economic NOUN bracis-19054 18 4 learning learn VERB bracis-19054 18 5 algorithms algorithm NOUN bracis-19054 18 6 machine machine NOUN bracis-19054 18 7 learning learn VERB bracis-19054 18 8 probabilistic probabilistic ADJ bracis-19054 18 9 data datum NOUN bracis-19054 18 10 networks network NOUN bracis-19054 18 11 self self NOUN bracis-19054 18 12 - - PUNCT bracis-19054 18 13 serving serve VERB bracis-19054 18 14 bias bias NOUN bracis-19054 18 15 1 1 NUM bracis-19054 18 16 introduction introduction NOUN bracis-19054 18 17 detecting detect VERB bracis-19054 18 18 fake fake ADJ bracis-19054 18 19 news news NOUN bracis-19054 18 20 is be AUX bracis-19054 18 21 a a DET bracis-19054 18 22 challenging challenging ADJ bracis-19054 18 23 task task NOUN bracis-19054 18 24 since since SCONJ bracis-19054 18 25 fake fake ADJ bracis-19054 18 26 news news NOUN bracis-19054 18 27 constantly constantly ADV bracis-19054 18 28 evolves evolve VERB bracis-19054 18 29 , , PUNCT bracis-19054 18 30 influencing influence VERB bracis-19054 18 31 the the DET bracis-19054 18 32 formation formation NOUN bracis-19054 18 33 of of ADP bracis-19054 18 34 the the DET bracis-19054 18 35 opinion opinion NOUN bracis-19054 18 36 of of ADP bracis-19054 18 37 social social ADJ bracis-19054 18 38 groups group NOUN bracis-19054 18 39 as as SCONJ bracis-19054 18 40 accepted accept VERB bracis-19054 18 41 [ [ X bracis-19054 18 42 6 6 NUM bracis-19054 18 43 ] ] PUNCT bracis-19054 18 44 . . PUNCT bracis-19054 19 1 to to PART bracis-19054 19 2 minimize minimize VERB bracis-19054 19 3 the the DET bracis-19054 19 4 effects effect NOUN bracis-19054 19 5 of of ADP bracis-19054 19 6 disinformation disinformation NOUN bracis-19054 19 7 provided provide VERB bracis-19054 19 8 by by ADP bracis-19054 19 9 fake fake ADJ bracis-19054 19 10 news news NOUN bracis-19054 19 11 , , PUNCT bracis-19054 19 12 machine machine NOUN bracis-19054 19 13 learning learning NOUN bracis-19054 19 14 algorithms algorithm NOUN bracis-19054 19 15 have have AUX bracis-19054 19 16 been be AUX bracis-19054 19 17 proposed propose VERB bracis-19054 19 18 in in ADP bracis-19054 19 19 the the DET bracis-19054 19 20 literature literature NOUN bracis-19054 19 21 , , PUNCT bracis-19054 19 22 which which PRON bracis-19054 19 23 learn learn VERB bracis-19054 19 24 classification classification NOUN bracis-19054 19 25 models model NOUN bracis-19054 19 26 to to PART bracis-19054 19 27 discriminate discriminate VERB bracis-19054 19 28 true true ADJ bracis-19054 19 29 and and CCONJ bracis-19054 19 30 fake fake ADJ bracis-19054 19 31 content content NOUN bracis-19054 19 32 [ [ X bracis-19054 19 33 4 4 NUM bracis-19054 19 34 , , PUNCT bracis-19054 19 35 22 22 NUM bracis-19054 19 36 ] ] PUNCT bracis-19054 19 37 . . PUNCT bracis-19054 20 1 the the DET bracis-19054 20 2 most most ADV bracis-19054 20 3 common common ADJ bracis-19054 20 4 way way NOUN bracis-19054 20 5 to to PART bracis-19054 20 6 deal deal VERB bracis-19054 20 7 with with ADP bracis-19054 20 8 this this DET bracis-19054 20 9 problem problem NOUN bracis-19054 20 10 is be AUX bracis-19054 20 11 to to PART bracis-19054 20 12 characterize characterize VERB bracis-19054 20 13 the the DET bracis-19054 20 14 fake fake ADJ bracis-19054 20 15 news news NOUN bracis-19054 20 16 detection detection NOUN bracis-19054 20 17 as as ADP bracis-19054 20 18 a a DET bracis-19054 20 19 supervised supervised ADJ bracis-19054 20 20 binary binary ADJ bracis-19054 20 21 or or CCONJ bracis-19054 20 22 multiclass multiclass ADJ bracis-19054 20 23 classification classification NOUN bracis-19054 20 24 problem problem NOUN bracis-19054 20 25 [ [ X bracis-19054 20 26 22 22 NUM bracis-19054 20 27 ] ] PUNCT bracis-19054 20 28 . . PUNCT bracis-19054 21 1 however however ADV bracis-19054 21 2 , , PUNCT bracis-19054 21 3 labeling label VERB bracis-19054 21 4 real real ADJ bracis-19054 21 5 and and CCONJ bracis-19054 21 6 fake fake ADJ bracis-19054 21 7 news news NOUN bracis-19054 21 8 to to PART bracis-19054 21 9 cover cover VERB bracis-19054 21 10 different different ADJ bracis-19054 21 11 subjects subject NOUN bracis-19054 21 12 , , PUNCT bracis-19054 21 13 sources source NOUN bracis-19054 21 14 , , PUNCT bracis-19054 21 15 and and CCONJ bracis-19054 21 16 falsity falsity NOUN bracis-19054 21 17 levels level NOUN bracis-19054 21 18 , , PUNCT bracis-19054 21 19 as as ADV bracis-19054 21 20 well well ADV bracis-19054 21 21 as as ADP bracis-19054 21 22 updating update VERB bracis-19054 21 23 the the DET bracis-19054 21 24 classification classification NOUN bracis-19054 21 25 model model NOUN bracis-19054 21 26 are be AUX bracis-19054 21 27 costly costly ADJ bracis-19054 21 28 processes process NOUN bracis-19054 21 29 [ [ X bracis-19054 21 30 4 4 NUM bracis-19054 21 31 ] ] PUNCT bracis-19054 21 32 . . PUNCT bracis-19054 22 1 all all PRON bracis-19054 22 2 of of ADP bracis-19054 22 3 these these DET bracis-19054 22 4 challenges challenge NOUN bracis-19054 22 5 drive drive VERB bracis-19054 22 6 the the DET bracis-19054 22 7 investigation investigation NOUN bracis-19054 22 8 of of ADP bracis-19054 22 9 more more ADV bracis-19054 22 10 appropriate appropriate ADJ bracis-19054 22 11 models model NOUN bracis-19054 22 12 for for ADP bracis-19054 22 13 representing represent VERB bracis-19054 22 14 and and CCONJ bracis-19054 22 15 classifying classify VERB bracis-19054 22 16 news news NOUN bracis-19054 22 17 . . PUNCT bracis-19054 23 1 positive positive ADJ bracis-19054 23 2 and and CCONJ bracis-19054 23 3 unlabeled unlabeled ADJ bracis-19054 23 4 learning learning NOUN bracis-19054 23 5 ( ( PUNCT bracis-19054 23 6 pul pul NOUN bracis-19054 23 7 ) ) PUNCT bracis-19054 23 8 algorithms algorithm NOUN bracis-19054 23 9 can can AUX bracis-19054 23 10 be be AUX bracis-19054 23 11 good good ADJ bracis-19054 23 12 alternatives alternative NOUN bracis-19054 23 13 in in ADP bracis-19054 23 14 this this DET bracis-19054 23 15 scenario scenario NOUN bracis-19054 23 16 [ [ X bracis-19054 23 17 3 3 NUM bracis-19054 23 18 ] ] PUNCT bracis-19054 23 19 . . PUNCT bracis-19054 24 1 pul pul NOUN bracis-19054 24 2 algorithms algorithm NOUN bracis-19054 24 3 learn learn VERB bracis-19054 24 4 models model NOUN bracis-19054 24 5 considering consider VERB bracis-19054 24 6 little little ADJ bracis-19054 24 7 labeled label VERB bracis-19054 24 8 data datum NOUN bracis-19054 24 9 of of ADP bracis-19054 24 10 the the DET bracis-19054 24 11 interest interest NOUN bracis-19054 24 12 class class NOUN bracis-19054 24 13 and and CCONJ bracis-19054 24 14 use use VERB bracis-19054 24 15 the the DET bracis-19054 24 16 unlabeled unlabeled ADJ bracis-19054 24 17 data datum NOUN bracis-19054 24 18 to to PART bracis-19054 24 19 increase increase VERB bracis-19054 24 20 classification classification NOUN bracis-19054 24 21 performance performance NOUN bracis-19054 24 22 [ [ X bracis-19054 24 23 9 9 NUM bracis-19054 24 24 ] ] PUNCT bracis-19054 24 25 . . PUNCT bracis-19054 25 1 therefore therefore ADV bracis-19054 25 2 , , PUNCT bracis-19054 25 3 pul pul NOUN bracis-19054 25 4 eliminates eliminate VERB bracis-19054 25 5 the the DET bracis-19054 25 6 need need NOUN bracis-19054 25 7 to to PART bracis-19054 25 8 label label VERB bracis-19054 25 9 a a DET bracis-19054 25 10 large large ADJ bracis-19054 25 11 number number NOUN bracis-19054 25 12 of of ADP bracis-19054 25 13 news news NOUN bracis-19054 25 14 of of ADP bracis-19054 25 15 the the DET bracis-19054 25 16 uninteresting uninteresting ADJ bracis-19054 25 17 classes class NOUN bracis-19054 25 18 . . PUNCT bracis-19054 26 1 due due ADP bracis-19054 26 2 to to ADP bracis-19054 26 3 the the DET bracis-19054 26 4 suitability suitability NOUN bracis-19054 26 5 of of ADP bracis-19054 26 6 pul pul NOUN bracis-19054 26 7 to to PART bracis-19054 26 8 fake fake VERB bracis-19054 26 9 news news NOUN bracis-19054 26 10 detection detection NOUN bracis-19054 26 11 scenario scenario NOUN bracis-19054 26 12 , , PUNCT bracis-19054 26 13 in in ADP bracis-19054 26 14 this this DET bracis-19054 26 15 paper paper NOUN bracis-19054 26 16 , , PUNCT bracis-19054 26 17 we we PRON bracis-19054 26 18 proposed propose VERB bracis-19054 26 19 an an DET bracis-19054 26 20 approach approach NOUN bracis-19054 26 21 based base VERB bracis-19054 26 22 on on ADP bracis-19054 26 23 the the DET bracis-19054 26 24 algorithm algorithm NOUN bracis-19054 26 25 positive positive ADJ bracis-19054 26 26 and and CCONJ bracis-19054 26 27 unlabeled unlabeled ADJ bracis-19054 26 28 learning learning NOUN bracis-19054 26 29 by by ADP bracis-19054 26 30 label label NOUN bracis-19054 26 31 propagation propagation NOUN bracis-19054 26 32 [ [ X bracis-19054 26 33 9 9 NUM bracis-19054 26 34 ] ] X bracis-19054 26 35 ( ( PUNCT bracis-19054 26 36 pu pu PROPN bracis-19054 26 37 - - PUNCT bracis-19054 26 38 lp lp NOUN bracis-19054 26 39 ) ) PUNCT bracis-19054 26 40 , , PUNCT bracis-19054 26 41 a a DET bracis-19054 26 42 network network NOUN bracis-19054 26 43 - - PUNCT bracis-19054 26 44 based base VERB bracis-19054 26 45 semi semi ADJ bracis-19054 26 46 - - ADJ bracis-19054 26 47 supervised supervised ADJ bracis-19054 26 48 transductive transductive ADJ bracis-19054 26 49 learning learn VERB bracis-19054 26 50 algorithm algorithm NOUN bracis-19054 26 51 . . PUNCT bracis-19054 27 1 pu pu PROPN bracis-19054 27 2 - - PUNCT bracis-19054 27 3 lp lp PROPN bracis-19054 27 4 has have AUX bracis-19054 27 5 been be AUX bracis-19054 27 6 little little ADV bracis-19054 27 7 explored explore VERB bracis-19054 27 8 for for ADP bracis-19054 27 9 text text NOUN bracis-19054 27 10 classification classification NOUN bracis-19054 27 11 , , PUNCT bracis-19054 27 12 despite despite SCONJ bracis-19054 27 13 its its PRON bracis-19054 27 14 good good ADJ bracis-19054 27 15 performance performance NOUN bracis-19054 27 16 in in ADP bracis-19054 27 17 numerical numerical ADJ bracis-19054 27 18 datasets dataset NOUN bracis-19054 27 19 . . PUNCT bracis-19054 28 1 pu pu PROPN bracis-19054 28 2 - - PUNCT bracis-19054 28 3 lp lp PROPN bracis-19054 28 4 infers infer NOUN bracis-19054 28 5 reliable reliable ADJ bracis-19054 28 6 sets set NOUN bracis-19054 28 7 of of ADP bracis-19054 28 8 interest interest NOUN bracis-19054 28 9 ( ( PUNCT bracis-19054 28 10 fake fake ADJ bracis-19054 28 11 news news NOUN bracis-19054 28 12 ) ) PUNCT bracis-19054 28 13 and and CCONJ bracis-19054 28 14 not not PART bracis-19054 28 15 interest interest NOUN bracis-19054 28 16 ( ( PUNCT bracis-19054 28 17 true true ADJ bracis-19054 28 18 news news NOUN bracis-19054 28 19 ) ) PUNCT bracis-19054 28 20 using use VERB bracis-19054 28 21 a a DET bracis-19054 28 22 small small ADJ bracis-19054 28 23 number number NOUN bracis-19054 28 24 of of ADP bracis-19054 28 25 labeled label VERB bracis-19054 28 26 data data PROPN bracis-19054 28 27 , , PUNCT bracis-19054 28 28 katz katz PROPN bracis-19054 28 29 index index PROPN bracis-19054 28 30 and and CCONJ bracis-19054 28 31 a a DET bracis-19054 28 32 k k NOUN bracis-19054 28 33 - - PUNCT bracis-19054 28 34 nearest near ADJ bracis-19054 28 35 neighbor neighbor NOUN bracis-19054 28 36 ( ( PUNCT bracis-19054 28 37 k k PROPN bracis-19054 28 38 - - PUNCT bracis-19054 28 39 nn nn ADJ bracis-19054 28 40 ) ) PUNCT bracis-19054 28 41 network network NOUN bracis-19054 28 42 . . PUNCT bracis-19054 29 1 then then ADV bracis-19054 29 2 , , PUNCT bracis-19054 29 3 a a DET bracis-19054 29 4 label label NOUN bracis-19054 29 5 propagation propagation NOUN bracis-19054 29 6 algorithm algorithm NOUN bracis-19054 29 7 is be AUX bracis-19054 29 8 used use VERB bracis-19054 29 9 to to PART bracis-19054 29 10 label label VERB bracis-19054 29 11 the the DET bracis-19054 29 12 remaining remain VERB bracis-19054 29 13 unlabeled unlabeled ADJ bracis-19054 29 14 objects object NOUN bracis-19054 29 15 as as ADP bracis-19054 29 16 interest interest NOUN bracis-19054 29 17 or or CCONJ bracis-19054 29 18 not not PART bracis-19054 29 19 interest interest NOUN bracis-19054 29 20 . . PUNCT bracis-19054 30 1 representing represent VERB bracis-19054 30 2 news news NOUN bracis-19054 30 3 using use VERB bracis-19054 30 4 traditional traditional ADJ bracis-19054 30 5 models model NOUN bracis-19054 30 6 , , PUNCT bracis-19054 30 7 i.e. i.e. X bracis-19054 30 8 , , PUNCT bracis-19054 30 9 bag bag NOUN bracis-19054 30 10 - - PUNCT bracis-19054 30 11 of of ADP bracis-19054 30 12 - - PUNCT bracis-19054 30 13 words word NOUN bracis-19054 30 14 or or CCONJ bracis-19054 30 15 embeddings embedding NOUN bracis-19054 30 16 , , PUNCT bracis-19054 30 17 may may AUX bracis-19054 30 18 not not PART bracis-19054 30 19 be be AUX bracis-19054 30 20 discriminative discriminative NOUN bracis-19054 30 21 enough enough ADV bracis-19054 30 22 to to PART bracis-19054 30 23 distinguish distinguish VERB bracis-19054 30 24 fake fake ADJ bracis-19054 30 25 and and CCONJ bracis-19054 30 26 real real ADJ bracis-19054 30 27 content content NOUN bracis-19054 30 28 [ [ X bracis-19054 30 29 17 17 NUM bracis-19054 30 30 , , PUNCT bracis-19054 30 31 21 21 NUM bracis-19054 30 32 ] ] PUNCT bracis-19054 30 33 . . PUNCT bracis-19054 31 1 thus thus ADV bracis-19054 31 2 , , PUNCT bracis-19054 31 3 an an DET bracis-19054 31 4 important important ADJ bracis-19054 31 5 challenge challenge NOUN bracis-19054 31 6 for for ADP bracis-19054 31 7 automatic automatic ADJ bracis-19054 31 8 fake fake ADJ bracis-19054 31 9 news news NOUN bracis-19054 31 10 detection detection NOUN bracis-19054 31 11 is be AUX bracis-19054 31 12 to to PART bracis-19054 31 13 assemble assemble VERB bracis-19054 31 14 a a DET bracis-19054 31 15 set set NOUN bracis-19054 31 16 of of ADP bracis-19054 31 17 features feature NOUN bracis-19054 31 18 that that PRON bracis-19054 31 19 efficiently efficiently ADV bracis-19054 31 20 characterize characterize VERB bracis-19054 31 21 a a DET bracis-19054 31 22 false false ADJ bracis-19054 31 23 content content NOUN bracis-19054 31 24 in in ADP bracis-19054 31 25 a a DET bracis-19054 31 26 dynamic dynamic ADJ bracis-19054 31 27 scenario scenario NOUN bracis-19054 31 28 . . PUNCT bracis-19054 32 1 this this DET bracis-19054 32 2 set set NOUN bracis-19054 32 3 may may AUX bracis-19054 32 4 contain contain VERB bracis-19054 32 5 features feature NOUN bracis-19054 32 6 about about ADP bracis-19054 32 7 creator creator NOUN bracis-19054 32 8 or or CCONJ bracis-19054 32 9 spreader spreader NOUN bracis-19054 32 10 , , PUNCT bracis-19054 32 11 the the DET bracis-19054 32 12 target target NOUN bracis-19054 32 13 victims victim NOUN bracis-19054 32 14 , , PUNCT bracis-19054 32 15 the the DET bracis-19054 32 16 news news NOUN bracis-19054 32 17 content content NOUN bracis-19054 32 18 and and CCONJ bracis-19054 32 19 the the DET bracis-19054 32 20 social social ADJ bracis-19054 32 21 context context NOUN bracis-19054 33 1 [ [ X bracis-19054 33 2 22 22 NUM bracis-19054 33 3 ] ] PUNCT bracis-19054 33 4 . . PUNCT bracis-19054 34 1 in in ADP bracis-19054 34 2 this this DET bracis-19054 34 3 work work NOUN bracis-19054 34 4 , , PUNCT bracis-19054 34 5 we we PRON bracis-19054 34 6 propose propose VERB bracis-19054 34 7 using use VERB bracis-19054 34 8 content content NOUN bracis-19054 34 9 - - PUNCT bracis-19054 34 10 based base VERB bracis-19054 34 11 features feature NOUN bracis-19054 34 12 in in ADP bracis-19054 34 13 a a DET bracis-19054 34 14 domain domain NOUN bracis-19054 34 15 - - PUNCT bracis-19054 34 16 independent independent ADJ bracis-19054 34 17 fake fake ADJ bracis-19054 34 18 news news NOUN bracis-19054 34 19 detection detection NOUN bracis-19054 34 20 approach approach NOUN bracis-19054 34 21 based base VERB bracis-19054 34 22 on on ADP bracis-19054 34 23 the the DET bracis-19054 34 24 pu pu PROPN bracis-19054 34 25 - - PUNCT bracis-19054 34 26 lp lp PROPN bracis-19054 34 27 algorithm algorithm NOUN bracis-19054 34 28 . . PUNCT bracis-19054 35 1 our our PRON bracis-19054 35 2 network network NOUN bracis-19054 35 3 use use VERB bracis-19054 35 4 linguistic linguistic ADJ bracis-19054 35 5 features feature NOUN bracis-19054 35 6 such such ADJ bracis-19054 35 7 as as ADP bracis-19054 35 8 representative representative ADJ bracis-19054 35 9 terms term NOUN bracis-19054 35 10 , , PUNCT bracis-19054 35 11 news news NOUN bracis-19054 35 12 emotiveness emotiveness NOUN bracis-19054 35 13 , , PUNCT bracis-19054 35 14 average average ADJ bracis-19054 35 15 number number NOUN bracis-19054 35 16 of of ADP bracis-19054 35 17 words word NOUN bracis-19054 35 18 per per ADP bracis-19054 35 19 sentence sentence NOUN bracis-19054 35 20 , , PUNCT bracis-19054 35 21 and and CCONJ bracis-19054 35 22 pausality pausality NOUN bracis-19054 35 23 as as ADP bracis-19054 35 24 network network NOUN bracis-19054 35 25 objects object NOUN bracis-19054 35 26 , , PUNCT bracis-19054 35 27 thus thus ADV bracis-19054 35 28 making make VERB bracis-19054 35 29 a a DET bracis-19054 35 30 heterogeneous heterogeneous ADJ bracis-19054 35 31 network network NOUN bracis-19054 35 32 [ [ X bracis-19054 35 33 12 12 NUM bracis-19054 35 34 ] ] PUNCT bracis-19054 35 35 . . PUNCT bracis-19054 36 1 these these DET bracis-19054 36 2 features feature NOUN bracis-19054 36 3 have have AUX bracis-19054 36 4 proven prove VERB bracis-19054 36 5 to to PART bracis-19054 36 6 be be AUX bracis-19054 36 7 relevant relevant ADJ bracis-19054 36 8 for for ADP bracis-19054 36 9 news news NOUN bracis-19054 36 10 classification classification NOUN bracis-19054 36 11 [ [ X bracis-19054 36 12 13 13 NUM bracis-19054 36 13 ] ] PUNCT bracis-19054 36 14 . . PUNCT bracis-19054 37 1 also also ADV bracis-19054 37 2 , , PUNCT bracis-19054 37 3 the the DET bracis-19054 37 4 presence presence NOUN bracis-19054 37 5 of of ADP bracis-19054 37 6 different different ADJ bracis-19054 37 7 objects object NOUN bracis-19054 37 8 and and CCONJ bracis-19054 37 9 relations relation NOUN bracis-19054 37 10 in in ADP bracis-19054 37 11 the the DET bracis-19054 37 12 network network NOUN bracis-19054 37 13 allows allow VERB bracis-19054 37 14 for for ADP bracis-19054 37 15 representing represent VERB bracis-19054 37 16 different different ADJ bracis-19054 37 17 patterns pattern NOUN bracis-19054 37 18 on on ADP bracis-19054 37 19 data datum NOUN bracis-19054 37 20 [ [ X bracis-19054 37 21 15 15 NUM bracis-19054 37 22 ] ] PUNCT bracis-19054 37 23 . . PUNCT bracis-19054 38 1 to to PART bracis-19054 38 2 create create VERB bracis-19054 38 3 the the DET bracis-19054 38 4 k k PROPN bracis-19054 38 5 - - PUNCT bracis-19054 38 6 nn nn ADJ bracis-19054 38 7 news news NOUN bracis-19054 38 8 network network NOUN bracis-19054 38 9 , , PUNCT bracis-19054 38 10 we we PRON bracis-19054 38 11 represent represent VERB bracis-19054 38 12 the the DET bracis-19054 38 13 text text NOUN bracis-19054 38 14 with with ADP bracis-19054 38 15 bag bag NOUN bracis-19054 38 16 - - PUNCT bracis-19054 38 17 of of ADP bracis-19054 38 18 - - PUNCT bracis-19054 38 19 words word NOUN bracis-19054 38 20 ( ( PUNCT bracis-19054 38 21 bow bow NOUN bracis-19054 38 22 ) ) PUNCT bracis-19054 38 23 and and CCONJ bracis-19054 38 24 doc2vec doc2vec PROPN bracis-19054 38 25 ( ( PUNCT bracis-19054 38 26 d2v d2v NOUN bracis-19054 38 27 ) ) PUNCT bracis-19054 38 28 models model NOUN bracis-19054 38 29 to to PART bracis-19054 38 30 compute compute VERB bracis-19054 38 31 the the DET bracis-19054 38 32 similarity similarity NOUN bracis-19054 38 33 between between ADP bracis-19054 38 34 the the DET bracis-19054 38 35 news news NOUN bracis-19054 38 36 . . PUNCT bracis-19054 39 1 to to ADP bracis-19054 39 2 the the DET bracis-19054 39 3 best good ADJ bracis-19054 39 4 of of ADP bracis-19054 39 5 our our PRON bracis-19054 39 6 knowledge knowledge NOUN bracis-19054 39 7 , , PUNCT bracis-19054 39 8 heterogeneous heterogeneous ADJ bracis-19054 39 9 networks network NOUN bracis-19054 39 10 and and CCONJ bracis-19054 39 11 pul pul NOUN bracis-19054 39 12 have have AUX bracis-19054 39 13 not not PART bracis-19054 39 14 been be AUX bracis-19054 39 15 explored explore VERB bracis-19054 39 16 in in ADP bracis-19054 39 17 the the DET bracis-19054 39 18 literature literature NOUN bracis-19054 39 19 to to PART bracis-19054 39 20 detect detect VERB bracis-19054 39 21 fake fake ADJ bracis-19054 39 22 news news NOUN bracis-19054 39 23 . . PUNCT bracis-19054 40 1 we we PRON bracis-19054 40 2 evaluated evaluate VERB bracis-19054 40 3 different different ADJ bracis-19054 40 4 combinations combination NOUN bracis-19054 40 5 of of ADP bracis-19054 40 6 features feature NOUN bracis-19054 40 7 to to PART bracis-19054 40 8 measure measure VERB bracis-19054 40 9 the the DET bracis-19054 40 10 impact impact NOUN bracis-19054 40 11 of of ADP bracis-19054 40 12 these these DET bracis-19054 40 13 features feature NOUN bracis-19054 40 14 in in ADP bracis-19054 40 15 the the DET bracis-19054 40 16 classification classification NOUN bracis-19054 40 17 performance performance NOUN bracis-19054 40 18 . . PUNCT bracis-19054 41 1 our our PRON bracis-19054 41 2 experiments experiment NOUN bracis-19054 41 3 used use VERB bracis-19054 41 4 relatively relatively ADV bracis-19054 41 5 small small ADJ bracis-19054 41 6 sets set NOUN bracis-19054 41 7 of of ADP bracis-19054 41 8 labeled label VERB bracis-19054 41 9 fake fake ADJ bracis-19054 41 10 news news NOUN bracis-19054 41 11 : : PUNCT bracis-19054 41 12 from from ADP bracis-19054 41 13 10 10 NUM bracis-19054 41 14 % % NOUN bracis-19054 41 15 to to PART bracis-19054 41 16 30 30 NUM bracis-19054 41 17 % % NOUN bracis-19054 41 18 of of ADP bracis-19054 41 19 the the DET bracis-19054 41 20 fake fake ADJ bracis-19054 41 21 news news NOUN bracis-19054 41 22 set set NOUN bracis-19054 41 23 . . PUNCT bracis-19054 42 1 the the DET bracis-19054 42 2 heterogeneous heterogeneous ADJ bracis-19054 42 3 network network NOUN bracis-19054 42 4 was be AUX bracis-19054 42 5 compared compare VERB bracis-19054 42 6 with with ADP bracis-19054 42 7 the the DET bracis-19054 42 8 k k PROPN bracis-19054 42 9 - - PUNCT bracis-19054 42 10 nn nn PROPN bracis-19054 42 11 networks network NOUN bracis-19054 42 12 , , PUNCT bracis-19054 42 13 used use VERB bracis-19054 42 14 in in ADP bracis-19054 42 15 the the DET bracis-19054 42 16 original original ADJ bracis-19054 42 17 proposal proposal NOUN bracis-19054 42 18 of of ADP bracis-19054 42 19 pu pu PROPN bracis-19054 42 20 - - PUNCT bracis-19054 42 21 lp lp PROPN bracis-19054 42 22 . . PUNCT bracis-19054 43 1 two two NUM bracis-19054 43 2 portuguese portuguese ADJ bracis-19054 43 3 datasets dataset NOUN bracis-19054 43 4 and and CCONJ bracis-19054 43 5 four four NUM bracis-19054 43 6 english english ADJ bracis-19054 43 7 datasets dataset NOUN bracis-19054 43 8 were be AUX bracis-19054 43 9 used use VERB bracis-19054 43 10 , , PUNCT bracis-19054 43 11 containing contain VERB bracis-19054 43 12 only only ADV bracis-19054 43 13 one one NUM bracis-19054 43 14 or or CCONJ bracis-19054 43 15 more more ADJ bracis-19054 43 16 subjects subject NOUN bracis-19054 43 17 . . PUNCT bracis-19054 44 1 for for ADP bracis-19054 44 2 the the DET bracis-19054 44 3 label label NOUN bracis-19054 44 4 propagation propagation NOUN bracis-19054 44 5 , , PUNCT bracis-19054 44 6 we we PRON bracis-19054 44 7 used use VERB bracis-19054 44 8 two two NUM bracis-19054 44 9 well well ADV bracis-19054 44 10 - - PUNCT bracis-19054 44 11 established establish VERB bracis-19054 44 12 algorithms algorithm NOUN bracis-19054 44 13 : : PUNCT bracis-19054 44 14 label label NOUN bracis-19054 44 15 propagation propagation NOUN bracis-19054 44 16 through through ADP bracis-19054 44 17 heterogeneous heterogeneous ADJ bracis-19054 44 18 networks network NOUN bracis-19054 44 19     SPACE bracis-19054 45 1 [ [ X bracis-19054 45 2 12 12 NUM bracis-19054 45 3 ] ] PUNCT bracis-19054 45 4 and and CCONJ bracis-19054 45 5 gnetmine gnetmine ADJ bracis-19054 45 6     SPACE bracis-19054 46 1 [ [ X bracis-19054 46 2 7 7 NUM bracis-19054 46 3 ] ] PUNCT bracis-19054 46 4 . . PUNCT bracis-19054 47 1 we we PRON bracis-19054 47 2 evaluated evaluate VERB bracis-19054 47 3 the the DET bracis-19054 47 4 algorithms algorithm NOUN bracis-19054 47 5 considering consider VERB bracis-19054 47 6 the the DET bracis-19054 47 7 \(f_{1}\ \(f_{1}\ ADJ bracis-19054 47 8 ) ) PUNCT bracis-19054 47 9 measure measure NOUN bracis-19054 47 10 of of ADP bracis-19054 47 11 the the DET bracis-19054 47 12 fake fake ADJ bracis-19054 47 13 news news NOUN bracis-19054 47 14 class class NOUN bracis-19054 47 15 . . PUNCT bracis-19054 48 1 the the DET bracis-19054 48 2 results result NOUN bracis-19054 48 3 obtained obtain VERB bracis-19054 48 4 the the DET bracis-19054 48 5 proposed propose VERB bracis-19054 48 6 approach approach NOUN bracis-19054 48 7 were be AUX bracis-19054 48 8 compared compare VERB bracis-19054 48 9 with with ADP bracis-19054 48 10 a a DET bracis-19054 48 11 semi semi ADJ bracis-19054 48 12 - - ADJ bracis-19054 48 13 supervised supervised ADJ bracis-19054 48 14 binary binary ADJ bracis-19054 48 15 baseline baseline NOUN bracis-19054 48 16 , , PUNCT bracis-19054 48 17 and and CCONJ bracis-19054 48 18 demonstrate demonstrate VERB bracis-19054 48 19 competitiveness competitiveness NOUN bracis-19054 48 20 even even ADV bracis-19054 48 21 with with ADP bracis-19054 48 22 half half DET bracis-19054 48 23 the the DET bracis-19054 48 24 news news NOUN bracis-19054 48 25 initially initially ADV bracis-19054 48 26 labeled label VERB bracis-19054 48 27 and and CCONJ bracis-19054 48 28 without without ADP bracis-19054 48 29 the the DET bracis-19054 48 30 labeling labeling NOUN bracis-19054 48 31 of of ADP bracis-19054 48 32 real real ADJ bracis-19054 48 33 news news NOUN bracis-19054 48 34 . . PUNCT bracis-19054 49 1 this this DET bracis-19054 49 2 paper paper NOUN bracis-19054 49 3 is be AUX bracis-19054 49 4 organized organize VERB bracis-19054 49 5 as as SCONJ bracis-19054 49 6 follows follow VERB bracis-19054 49 7 : : PUNCT bracis-19054 49 8 sects sect NOUN bracis-19054 49 9 . . PUNCT bracis-19054 49 10     SPACE bracis-19054 50 1 2 2 NUM bracis-19054 50 2 and and CCONJ bracis-19054 50 3 3 3 NUM bracis-19054 50 4 present present ADJ bracis-19054 50 5 related related ADJ bracis-19054 50 6 work work NOUN bracis-19054 50 7 . . PUNCT bracis-19054 51 1 section section NOUN bracis-19054 51 2     SPACE bracis-19054 51 3 4 4 NUM bracis-19054 51 4 describes describe VERB bracis-19054 51 5 the the DET bracis-19054 51 6 proposed propose VERB bracis-19054 51 7 approach approach NOUN bracis-19054 51 8 . . PUNCT bracis-19054 52 1 section section NOUN bracis-19054 52 2     SPACE bracis-19054 52 3 5 5 NUM bracis-19054 52 4 presents present VERB bracis-19054 52 5 experimental experimental ADJ bracis-19054 52 6 evaluation evaluation NOUN bracis-19054 52 7 and and CCONJ bracis-19054 52 8 discussions discussion NOUN bracis-19054 52 9 . . PUNCT bracis-19054 53 1 section section NOUN bracis-19054 53 2     SPACE bracis-19054 53 3 6 6 NUM bracis-19054 53 4 presents present VERB bracis-19054 53 5 conclusions conclusion NOUN bracis-19054 53 6 and and CCONJ bracis-19054 53 7 future future ADJ bracis-19054 53 8 works work NOUN bracis-19054 53 9 . . PUNCT bracis-19054 54 1 2 2 NUM bracis-19054 54 2 related relate VERB bracis-19054 54 3 work work NOUN bracis-19054 54 4 existing exist VERB bracis-19054 54 5 methods method NOUN bracis-19054 54 6 for for ADP bracis-19054 54 7 news news NOUN bracis-19054 54 8 classification classification NOUN bracis-19054 54 9 , , PUNCT bracis-19054 54 10 in in ADP bracis-19054 54 11 general general ADJ bracis-19054 54 12 , , PUNCT bracis-19054 54 13 represent represent VERB bracis-19054 54 14 news news NOUN bracis-19054 54 15 in in ADP bracis-19054 54 16 a a DET bracis-19054 54 17 structured structured ADJ bracis-19054 54 18 way way NOUN bracis-19054 54 19 using use VERB bracis-19054 54 20 traditional traditional ADJ bracis-19054 54 21 models model NOUN bracis-19054 54 22 such such ADJ bracis-19054 54 23 as as ADP bracis-19054 54 24 bow bow NOUN bracis-19054 54 25 , , PUNCT bracis-19054 54 26 word word NOUN bracis-19054 54 27 ( ( PUNCT bracis-19054 54 28 or or CCONJ bracis-19054 54 29 document document NOUN bracis-19054 54 30 ) ) PUNCT bracis-19054 54 31 embeddings embedding NOUN bracis-19054 54 32 or or CCONJ bracis-19054 54 33 networks network NOUN bracis-19054 54 34 [ [ X bracis-19054 54 35 22 22 NUM bracis-19054 54 36 ] ] PUNCT bracis-19054 54 37 . . PUNCT bracis-19054 55 1 however however ADV bracis-19054 55 2 , , PUNCT bracis-19054 55 3 these these DET bracis-19054 55 4 models model NOUN bracis-19054 55 5 are be AUX bracis-19054 55 6 not not PART bracis-19054 55 7 discriminative discriminative NOUN bracis-19054 55 8 enough enough ADV bracis-19054 55 9 to to PART bracis-19054 55 10 capture capture VERB bracis-19054 55 11 the the DET bracis-19054 55 12 nuances nuance NOUN bracis-19054 55 13 to to PART bracis-19054 55 14 distinguish distinguish VERB bracis-19054 55 15 fake fake ADJ bracis-19054 55 16 from from ADP bracis-19054 55 17 real real ADJ bracis-19054 55 18 news news NOUN bracis-19054 55 19 , , PUNCT bracis-19054 55 20 requiring require VERB bracis-19054 55 21 the the DET bracis-19054 55 22 investigation investigation NOUN bracis-19054 55 23 of of ADP bracis-19054 55 24 textual textual ADJ bracis-19054 55 25 and and CCONJ bracis-19054 55 26 contextual contextual ADJ bracis-19054 55 27 features feature NOUN bracis-19054 55 28 that that PRON bracis-19054 55 29 can can AUX bracis-19054 55 30 be be AUX bracis-19054 55 31 incorporated incorporate VERB bracis-19054 55 32 into into ADP bracis-19054 55 33 the the DET bracis-19054 55 34 representation representation NOUN bracis-19054 55 35 model model NOUN bracis-19054 55 36 [ [ X bracis-19054 55 37 17 17 NUM bracis-19054 55 38 , , PUNCT bracis-19054 55 39 21 21 NUM bracis-19054 55 40 , , PUNCT bracis-19054 55 41 22 22 NUM bracis-19054 55 42 ] ] PUNCT bracis-19054 55 43 . . PUNCT bracis-19054 56 1 contextual contextual ADJ bracis-19054 56 2 features feature NOUN bracis-19054 56 3 are be AUX bracis-19054 56 4 extracted extract VERB bracis-19054 56 5 from from ADP bracis-19054 56 6 the the DET bracis-19054 56 7 news news NOUN bracis-19054 56 8 environment environment NOUN bracis-19054 56 9 , , PUNCT bracis-19054 56 10 such such ADJ bracis-19054 56 11 as as ADP bracis-19054 56 12 post post ADJ bracis-19054 56 13 , , PUNCT bracis-19054 56 14 source source NOUN bracis-19054 56 15 , , PUNCT bracis-19054 56 16 and and CCONJ bracis-19054 56 17 publisher publisher NOUN bracis-19054 56 18 related relate VERB bracis-19054 56 19 information information NOUN bracis-19054 56 20 [ [ X bracis-19054 56 21 21 21 NUM bracis-19054 56 22 ] ] PUNCT bracis-19054 56 23 . . PUNCT bracis-19054 57 1 textual textual ADJ bracis-19054 57 2 features feature NOUN bracis-19054 57 3 involve involve VERB bracis-19054 57 4 lexical lexical ADJ bracis-19054 57 5 , , PUNCT bracis-19054 57 6 syntactic syntactic ADJ bracis-19054 57 7 and and CCONJ bracis-19054 57 8 semantic semantic ADJ bracis-19054 57 9 aspects aspect NOUN bracis-19054 57 10 of of ADP bracis-19054 57 11 the the DET bracis-19054 57 12 news news NOUN bracis-19054 57 13 content content NOUN bracis-19054 57 14 . . PUNCT bracis-19054 58 1 lexical lexical ADJ bracis-19054 58 2 features feature NOUN bracis-19054 58 3 are be AUX bracis-19054 58 4 related relate VERB bracis-19054 58 5 to to ADP bracis-19054 58 6 frequency frequency NOUN bracis-19054 58 7 statistics statistic NOUN bracis-19054 58 8 of of ADP bracis-19054 58 9 words word NOUN bracis-19054 58 10 , , PUNCT bracis-19054 58 11 which which PRON bracis-19054 58 12 can can AUX bracis-19054 58 13 be be AUX bracis-19054 58 14 done do VERB bracis-19054 58 15 using use VERB bracis-19054 58 16 n n CCONJ bracis-19054 58 17 - - PUNCT bracis-19054 58 18 gram gram NOUN bracis-19054 58 19 models model NOUN bracis-19054 58 20 , , PUNCT bracis-19054 58 21 able able ADJ bracis-19054 58 22 to to PART bracis-19054 58 23 identify identify VERB bracis-19054 58 24 , , PUNCT bracis-19054 58 25 for for ADP bracis-19054 58 26 example example NOUN bracis-19054 58 27 , , PUNCT bracis-19054 58 28 the the DET bracis-19054 58 29 presence presence NOUN bracis-19054 58 30 of of ADP bracis-19054 58 31 vague vague ADJ bracis-19054 58 32 , , PUNCT bracis-19054 58 33 doubt doubt NOUN bracis-19054 58 34 or or CCONJ bracis-19054 58 35 sensationalist sensationalist NOUN bracis-19054 58 36 expressions expression NOUN bracis-19054 58 37 [ [ X bracis-19054 58 38 4 4 NUM bracis-19054 58 39 ] ] PUNCT bracis-19054 58 40 . . PUNCT bracis-19054 59 1 syntactic syntactic ADJ bracis-19054 59 2 features feature NOUN bracis-19054 59 3 are be AUX bracis-19054 59 4 related relate VERB bracis-19054 59 5 to to ADP bracis-19054 59 6 the the DET bracis-19054 59 7 presence presence NOUN bracis-19054 59 8 and and CCONJ bracis-19054 59 9 frequency frequency NOUN bracis-19054 59 10 of of ADP bracis-19054 59 11 part part NOUN bracis-19054 59 12 - - PUNCT bracis-19054 59 13 of of ADP bracis-19054 59 14 - - PUNCT bracis-19054 59 15 speech speech NOUN bracis-19054 59 16 patterns pattern NOUN bracis-19054 59 17 , , PUNCT bracis-19054 59 18 like like ADP bracis-19054 59 19 subjects subject NOUN bracis-19054 59 20 , , PUNCT bracis-19054 59 21 verbs verbs PROPN bracis-19054 59 22 and and CCONJ bracis-19054 59 23 adjectives adjective NOUN bracis-19054 60 1 [ [ X bracis-19054 60 2 17 17 NUM bracis-19054 60 3 ] ] PUNCT bracis-19054 60 4 . . PUNCT bracis-19054 61 1 at at ADP bracis-19054 61 2 the the DET bracis-19054 61 3 semantic semantic ADJ bracis-19054 61 4 level level NOUN bracis-19054 61 5 , , PUNCT bracis-19054 61 6 meanings meaning NOUN bracis-19054 61 7 are be AUX bracis-19054 61 8 extracted extract VERB bracis-19054 61 9 from from ADP bracis-19054 61 10 terms term NOUN bracis-19054 61 11 , , PUNCT bracis-19054 61 12 where where SCONJ bracis-19054 61 13 methods method NOUN bracis-19054 61 14 like like ADP bracis-19054 61 15 linguistic linguistic ADJ bracis-19054 61 16 inquiry inquiry NOUN bracis-19054 61 17 and and CCONJ bracis-19054 61 18 word word NOUN bracis-19054 61 19 count count NOUN bracis-19054 61 20 ( ( PUNCT bracis-19054 61 21 liwc liwc PROPN bracis-19054 61 22 ) ) PUNCT bracis-19054 61 23     SPACE bracis-19054 62 1 [ [ X bracis-19054 62 2 10 10 NUM bracis-19054 62 3 ] ] PUNCT bracis-19054 62 4 can can AUX bracis-19054 62 5 be be AUX bracis-19054 62 6 used use VERB bracis-19054 62 7 to to PART bracis-19054 62 8 estimate estimate VERB bracis-19054 62 9 emotional emotional ADJ bracis-19054 62 10 , , PUNCT bracis-19054 62 11 cognitive cognitive ADJ bracis-19054 62 12 and and CCONJ bracis-19054 62 13 structural structural ADJ bracis-19054 62 14 components component NOUN bracis-19054 62 15 present present ADJ bracis-19054 62 16 in in ADP bracis-19054 62 17 the the DET bracis-19054 62 18 written write VERB bracis-19054 62 19 language language NOUN bracis-19054 62 20 . . PUNCT bracis-19054 63 1 in in ADP bracis-19054 63 2 [ [ X bracis-19054 63 3 17 17 NUM bracis-19054 63 4 ] ] PUNCT bracis-19054 63 5 , , PUNCT bracis-19054 63 6 authors author NOUN bracis-19054 63 7 proposed propose VERB bracis-19054 63 8 representation representation NOUN bracis-19054 63 9 models model NOUN bracis-19054 63 10 for for ADP bracis-19054 63 11 the the DET bracis-19054 63 12 fake.br fake.br PROPN bracis-19054 63 13 dataset dataset NOUN bracis-19054 63 14 that that PRON bracis-19054 63 15 combine combine AUX bracis-19054 63 16 bow bow VERB bracis-19054 63 17 and and CCONJ bracis-19054 63 18 linguistic linguistic ADJ bracis-19054 63 19 features feature NOUN bracis-19054 63 20 , , PUNCT bracis-19054 63 21 such such ADJ bracis-19054 63 22 as as ADP bracis-19054 63 23 pausality pausality NOUN bracis-19054 63 24 , , PUNCT bracis-19054 63 25 emotiveness emotiveness NOUN bracis-19054 63 26 , , PUNCT bracis-19054 63 27 and and CCONJ bracis-19054 63 28 uncertainty uncertainty NOUN bracis-19054 63 29 . . PUNCT bracis-19054 64 1 the the DET bracis-19054 64 2 authors author NOUN bracis-19054 64 3 reached reach VERB bracis-19054 64 4 96 96 NUM bracis-19054 64 5 % % NOUN bracis-19054 64 6 of of ADP bracis-19054 64 7 macro macro NOUN bracis-19054 64 8 \(f_{1}\ \(f_{1}\ NOUN bracis-19054 64 9 ) ) PUNCT bracis-19054 64 10 considering consider VERB bracis-19054 64 11 two two NUM bracis-19054 64 12 thirds third NOUN bracis-19054 64 13 of of ADP bracis-19054 64 14 the the DET bracis-19054 64 15 data datum NOUN bracis-19054 64 16 as as ADP bracis-19054 64 17 training training NOUN bracis-19054 64 18 . . PUNCT bracis-19054 65 1 in in ADP bracis-19054 65 2 [ [ X bracis-19054 65 3 18 18 NUM bracis-19054 65 4 ] ] PUNCT bracis-19054 65 5 , , PUNCT bracis-19054 65 6 the the DET bracis-19054 65 7 authors author NOUN bracis-19054 65 8 proposed propose VERB bracis-19054 65 9 a a DET bracis-19054 65 10 multimodal multimodal ADJ bracis-19054 65 11 representation representation NOUN bracis-19054 65 12 that that PRON bracis-19054 65 13 combines combine VERB bracis-19054 65 14 text text NOUN bracis-19054 65 15 and and CCONJ bracis-19054 65 16 images image VERB bracis-19054 65 17 information information NOUN bracis-19054 65 18 . . PUNCT bracis-19054 66 1 for for ADP bracis-19054 66 2 each each DET bracis-19054 66 3 news news NOUN bracis-19054 66 4 , , PUNCT bracis-19054 66 5 124 124 NUM bracis-19054 66 6 textual textual ADJ bracis-19054 66 7 features feature NOUN bracis-19054 66 8 ( ( PUNCT bracis-19054 66 9 81 81 NUM bracis-19054 66 10 of of ADP bracis-19054 66 11 them they PRON bracis-19054 66 12 with with ADP bracis-19054 66 13 liwc liwc PROPN bracis-19054 66 14 ) ) PUNCT bracis-19054 66 15 and and CCONJ bracis-19054 66 16 43 43 NUM bracis-19054 66 17 image image NOUN bracis-19054 66 18 features feature NOUN bracis-19054 66 19 are be AUX bracis-19054 66 20 collected collect VERB bracis-19054 66 21 . . PUNCT bracis-19054 67 1 the the DET bracis-19054 67 2 approach approach NOUN bracis-19054 67 3 achieves achieve VERB bracis-19054 67 4 95 95 NUM bracis-19054 67 5 % % NOUN bracis-19054 67 6 accuracy accuracy NOUN bracis-19054 67 7 with with ADP bracis-19054 67 8 random random ADJ bracis-19054 67 9 forest forest NOUN bracis-19054 67 10 using use VERB bracis-19054 67 11 approximately approximately ADV bracis-19054 67 12 7,000 7,000 NUM bracis-19054 67 13 news news NOUN bracis-19054 67 14 documents document NOUN bracis-19054 67 15 and and CCONJ bracis-19054 67 16 90 90 NUM bracis-19054 67 17 % % NOUN bracis-19054 67 18 of of ADP bracis-19054 67 19 data datum NOUN bracis-19054 67 20 in in ADP bracis-19054 67 21 the the DET bracis-19054 67 22 training training NOUN bracis-19054 67 23 set set NOUN bracis-19054 67 24 . . PUNCT bracis-19054 68 1 some some DET bracis-19054 68 2 works work VERB bracis-19054 68 3 use use VERB bracis-19054 68 4 networks network NOUN bracis-19054 68 5 as as ADP bracis-19054 68 6 a a DET bracis-19054 68 7 representation representation NOUN bracis-19054 68 8 model model NOUN bracis-19054 68 9 to to PART bracis-19054 68 10 detect detect VERB bracis-19054 68 11 false false ADJ bracis-19054 68 12 information information NOUN bracis-19054 68 13 . . PUNCT bracis-19054 69 1 [ [ X bracis-19054 69 2 13 13 NUM bracis-19054 69 3 ] ] PUNCT bracis-19054 69 4 proposes propose VERB bracis-19054 69 5 a a DET bracis-19054 69 6 knowledge knowledge NOUN bracis-19054 69 7 network network NOUN bracis-19054 69 8 for for ADP bracis-19054 69 9 fact fact NOUN bracis-19054 69 10 - - PUNCT bracis-19054 69 11 checking checking NOUN bracis-19054 69 12 , , PUNCT bracis-19054 69 13 in in ADP bracis-19054 69 14 which which PRON bracis-19054 69 15 the the DET bracis-19054 69 16 entry entry NOUN bracis-19054 69 17 is be AUX bracis-19054 69 18 a a DET bracis-19054 69 19 triple triple ADJ bracis-19054 69 20 ( ( PUNCT bracis-19054 69 21 subject subject ADJ bracis-19054 69 22 , , PUNCT bracis-19054 69 23 predicate predicate NOUN bracis-19054 69 24 , , PUNCT bracis-19054 69 25 object object NOUN bracis-19054 69 26 ) ) PUNCT bracis-19054 69 27 . . PUNCT bracis-19054 70 1 the the DET bracis-19054 70 2 network network NOUN bracis-19054 70 3 is be AUX bracis-19054 70 4 constructed construct VERB bracis-19054 70 5 with with ADP bracis-19054 70 6 information information NOUN bracis-19054 70 7 from from ADP bracis-19054 70 8 wikipedia wikipedia PROPN bracis-19054 70 9 , , PUNCT bracis-19054 70 10 in in ADP bracis-19054 70 11 which which PRON bracis-19054 70 12 given give VERB bracis-19054 70 13 a a DET bracis-19054 70 14 sentence sentence NOUN bracis-19054 70 15 , , PUNCT bracis-19054 70 16 if if SCONJ bracis-19054 70 17 it it PRON bracis-19054 70 18 exists exist VERB bracis-19054 70 19 within within ADP bracis-19054 70 20 the the DET bracis-19054 70 21 knowledge knowledge NOUN bracis-19054 70 22 network network NOUN bracis-19054 70 23 , , PUNCT bracis-19054 70 24 it it PRON bracis-19054 70 25 is be AUX bracis-19054 70 26 considered consider VERB bracis-19054 70 27 authentic authentic ADJ bracis-19054 70 28 . . PUNCT bracis-19054 71 1 the the DET bracis-19054 71 2 approach approach NOUN bracis-19054 71 3 achieves achieve VERB bracis-19054 71 4 approximately approximately ADV bracis-19054 71 5 74 74 NUM bracis-19054 71 6 % % NOUN bracis-19054 71 7 accuracy accuracy NOUN bracis-19054 71 8 . . PUNCT bracis-19054 72 1 [ [ X bracis-19054 72 2 21 21 NUM bracis-19054 72 3 ] ] PUNCT bracis-19054 72 4 proposes propose VERB bracis-19054 72 5 a a DET bracis-19054 72 6 diffuse diffuse ADJ bracis-19054 72 7 neural neural ADJ bracis-19054 72 8 network network NOUN bracis-19054 72 9 graph graph NOUN bracis-19054 72 10 for for ADP bracis-19054 72 11 inferring infer VERB bracis-19054 72 12 the the DET bracis-19054 72 13 credibility credibility NOUN bracis-19054 72 14 of of ADP bracis-19054 72 15 news news NOUN bracis-19054 72 16 , , PUNCT bracis-19054 72 17 creators creator NOUN bracis-19054 72 18 and and CCONJ bracis-19054 72 19 subjects subject NOUN bracis-19054 72 20 . . PUNCT bracis-19054 73 1 the the DET bracis-19054 73 2 graph graph NOUN bracis-19054 73 3 has have VERB bracis-19054 73 4 approximately approximately ADV bracis-19054 73 5 14,000 14,000 NUM bracis-19054 73 6 news news NOUN bracis-19054 73 7 items item NOUN bracis-19054 73 8 , , PUNCT bracis-19054 73 9 3,600 3,600 NUM bracis-19054 73 10 creators creator NOUN bracis-19054 73 11 and and CCONJ bracis-19054 73 12 152 152 NUM bracis-19054 73 13 subjects subject NOUN bracis-19054 73 14 . . PUNCT bracis-19054 74 1 the the DET bracis-19054 74 2 approach approach NOUN bracis-19054 74 3 achieves achieve VERB bracis-19054 74 4 the the DET bracis-19054 74 5 best good ADJ bracis-19054 74 6 overall overall ADJ bracis-19054 74 7 performance performance NOUN bracis-19054 74 8 in in ADP bracis-19054 74 9 the the DET bracis-19054 74 10 binary binary ADJ bracis-19054 74 11 classification classification NOUN bracis-19054 74 12 compared compare VERB bracis-19054 74 13 to to ADP bracis-19054 74 14 svm svm PROPN bracis-19054 74 15 , , PUNCT bracis-19054 74 16 deep deep ADJ bracis-19054 74 17 walk walk NOUN bracis-19054 74 18 and and CCONJ bracis-19054 74 19 a a DET bracis-19054 74 20 neural neural ADJ bracis-19054 74 21 network network NOUN bracis-19054 74 22 . . PUNCT bracis-19054 75 1 [ [ X bracis-19054 75 2 11 11 NUM bracis-19054 75 3 ] ] PUNCT bracis-19054 75 4 proposes propose VERB bracis-19054 75 5 an an DET bracis-19054 75 6 adversarial adversarial ADJ bracis-19054 75 7 active active ADJ bracis-19054 75 8 learning learn VERB bracis-19054 75 9 heterogeneous heterogeneous ADJ bracis-19054 75 10 graph graph NOUN bracis-19054 75 11 neural neural ADJ bracis-19054 75 12 network network NOUN bracis-19054 75 13 to to PART bracis-19054 75 14 detect detect VERB bracis-19054 75 15 fake fake ADJ bracis-19054 75 16 news news NOUN bracis-19054 75 17 . . PUNCT bracis-19054 76 1 a a DET bracis-19054 76 2 hierarchical hierarchical ADJ bracis-19054 76 3 attention attention NOUN bracis-19054 76 4 mechanism mechanism NOUN bracis-19054 76 5 is be AUX bracis-19054 76 6 used use VERB bracis-19054 76 7 to to PART bracis-19054 76 8 learn learn VERB bracis-19054 76 9 nodes node NOUN bracis-19054 76 10 ’ ' PUNCT bracis-19054 76 11 representations representation NOUN bracis-19054 76 12 , , PUNCT bracis-19054 76 13 and and CCONJ bracis-19054 76 14 a a DET bracis-19054 76 15 selector selector NOUN bracis-19054 76 16 is be AUX bracis-19054 76 17 responsible responsible ADJ bracis-19054 76 18 for for ADP bracis-19054 76 19 consulting consult VERB bracis-19054 76 20 high high ADJ bracis-19054 76 21 - - PUNCT bracis-19054 76 22 value value NOUN bracis-19054 76 23 candidates candidate NOUN bracis-19054 76 24 for for ADP bracis-19054 76 25 active active ADJ bracis-19054 76 26 learning learning NOUN bracis-19054 76 27 . . PUNCT bracis-19054 77 1 the the DET bracis-19054 77 2 authors author NOUN bracis-19054 77 3 evaluate evaluate VERB bracis-19054 77 4 two two NUM bracis-19054 77 5 datasets dataset NOUN bracis-19054 77 6 : : PUNCT bracis-19054 77 7 the the DET bracis-19054 77 8 first first ADV bracis-19054 77 9 containing contain VERB bracis-19054 77 10 14,055 14,055 NUM bracis-19054 77 11 news news NOUN bracis-19054 77 12 , , PUNCT bracis-19054 77 13 3,634 3,634 NUM bracis-19054 77 14 creators creator NOUN bracis-19054 77 15 and and CCONJ bracis-19054 77 16 152 152 NUM bracis-19054 77 17 subjects subject NOUN bracis-19054 77 18 , , PUNCT bracis-19054 77 19 and and CCONJ bracis-19054 77 20 the the DET bracis-19054 77 21 second second NOUN bracis-19054 77 22 containing contain VERB bracis-19054 77 23 182 182 NUM bracis-19054 77 24 news news NOUN bracis-19054 77 25 , , PUNCT bracis-19054 77 26 15,257 15,257 NUM bracis-19054 77 27 twitter twitter NOUN bracis-19054 77 28 users user NOUN bracis-19054 77 29 and and CCONJ bracis-19054 77 30 9 9 NUM bracis-19054 77 31 publishers publisher NOUN bracis-19054 77 32 . . PUNCT bracis-19054 78 1 with with ADP bracis-19054 78 2 20 20 NUM bracis-19054 78 3 % % NOUN bracis-19054 78 4 of of ADP bracis-19054 78 5 real real ADJ bracis-19054 78 6 and and CCONJ bracis-19054 78 7 fake fake ADJ bracis-19054 78 8 news news NOUN bracis-19054 78 9 labeled label VERB bracis-19054 78 10 , , PUNCT bracis-19054 78 11 the the DET bracis-19054 78 12 algorithm algorithm NOUN bracis-19054 78 13 reaches reach VERB bracis-19054 78 14 respectively respectively ADV bracis-19054 78 15 57 57 NUM bracis-19054 78 16 % % NOUN bracis-19054 78 17 and and CCONJ bracis-19054 78 18 70 70 NUM bracis-19054 78 19 % % NOUN bracis-19054 78 20 of of ADP bracis-19054 78 21 macro macro NOUN bracis-19054 78 22 \(f_{1}\ \(f_{1}\ NOUN bracis-19054 78 23 ) ) PUNCT bracis-19054 78 24 . . PUNCT bracis-19054 79 1 [ [ X bracis-19054 79 2 19 19 NUM bracis-19054 79 3 ] ] PUNCT bracis-19054 79 4 proposes propose VERB bracis-19054 79 5 iarnet iarnet NOUN bracis-19054 79 6 , , PUNCT bracis-19054 79 7 an an DET bracis-19054 79 8 information information NOUN bracis-19054 79 9 aggregating aggregate VERB bracis-19054 79 10 and and CCONJ bracis-19054 79 11 reasoning reason VERB bracis-19054 79 12 network network NOUN bracis-19054 79 13 to to PART bracis-19054 79 14 detect detect VERB bracis-19054 79 15 fake fake ADJ bracis-19054 79 16 news news NOUN bracis-19054 79 17 on on ADP bracis-19054 79 18 heterogeneous heterogeneous ADJ bracis-19054 79 19 networks network NOUN bracis-19054 79 20 . . PUNCT bracis-19054 80 1 the the DET bracis-19054 80 2 network network NOUN bracis-19054 80 3 contains contain VERB bracis-19054 80 4 the the DET bracis-19054 80 5 post post PROPN bracis-19054 80 6 source source NOUN bracis-19054 80 7 , , PUNCT bracis-19054 80 8 comments comment NOUN bracis-19054 80 9 and and CCONJ bracis-19054 80 10 users user NOUN bracis-19054 80 11 as as ADP bracis-19054 80 12 nodes node NOUN bracis-19054 80 13 and and CCONJ bracis-19054 80 14 interactions interaction NOUN bracis-19054 80 15 between between ADP bracis-19054 80 16 them they PRON bracis-19054 80 17 as as ADP bracis-19054 80 18 edges edge NOUN bracis-19054 80 19 . . PUNCT bracis-19054 81 1 the the DET bracis-19054 81 2 authors author NOUN bracis-19054 81 3 evaluated evaluate VERB bracis-19054 81 4 two two NUM bracis-19054 81 5 datasets dataset NOUN bracis-19054 81 6 , , PUNCT bracis-19054 81 7 weibo weibo NOUN bracis-19054 81 8 and and CCONJ bracis-19054 81 9 fakeddit fakeddit NOUN bracis-19054 81 10 , , PUNCT bracis-19054 81 11 and and CCONJ bracis-19054 81 12 70 70 NUM bracis-19054 81 13 % % NOUN bracis-19054 81 14 of of ADP bracis-19054 81 15 the the DET bracis-19054 81 16 labeled label VERB bracis-19054 81 17 news news NOUN bracis-19054 81 18 for for ADP bracis-19054 81 19 training training NOUN bracis-19054 81 20 . . PUNCT bracis-19054 82 1 the the DET bracis-19054 82 2 algorithm algorithm NOUN bracis-19054 82 3 achieved achieve VERB bracis-19054 82 4 96 96 NUM bracis-19054 82 5 % % NOUN bracis-19054 82 6 accuracy accuracy NOUN bracis-19054 82 7 . . PUNCT bracis-19054 83 1 to to PART bracis-19054 83 2 avoid avoid VERB bracis-19054 83 3 real real ADJ bracis-19054 83 4 news news NOUN bracis-19054 83 5 labeling labeling NOUN bracis-19054 83 6 efforts effort NOUN bracis-19054 83 7 , , PUNCT bracis-19054 83 8 [ [ X bracis-19054 83 9 5 5 NUM bracis-19054 83 10 ] ] PUNCT bracis-19054 83 11 proposes propose VERB bracis-19054 83 12 the the DET bracis-19054 83 13 dcdistanceocc dcdistanceocc NOUN bracis-19054 83 14 , , PUNCT bracis-19054 83 15 a a DET bracis-19054 83 16 one one NUM bracis-19054 83 17 - - PUNCT bracis-19054 83 18 class class NOUN bracis-19054 83 19 learning learning NOUN bracis-19054 83 20 ( ( PUNCT bracis-19054 83 21 ocl ocl PROPN bracis-19054 83 22 ) ) PUNCT bracis-19054 83 23 algorithm algorithm NOUN bracis-19054 83 24 to to PART bracis-19054 83 25 identify identify VERB bracis-19054 83 26 false false ADJ bracis-19054 83 27 content content NOUN bracis-19054 83 28 . . PUNCT bracis-19054 84 1 for for ADP bracis-19054 84 2 each each DET bracis-19054 84 3 news news NOUN bracis-19054 84 4 item item NOUN bracis-19054 84 5 , , PUNCT bracis-19054 84 6 linguistic linguistic ADJ bracis-19054 84 7 characteristics characteristic NOUN bracis-19054 84 8 are be AUX bracis-19054 84 9 extracted extract VERB bracis-19054 84 10 , , PUNCT bracis-19054 84 11 such such ADJ bracis-19054 84 12 as as ADP bracis-19054 84 13 number number NOUN bracis-19054 84 14 of of ADP bracis-19054 84 15 words word NOUN bracis-19054 84 16 per per ADP bracis-19054 84 17 sentence sentence NOUN bracis-19054 84 18 and and CCONJ bracis-19054 84 19 sentiment sentiment NOUN bracis-19054 84 20 of of ADP bracis-19054 84 21 the the DET bracis-19054 84 22 message message NOUN bracis-19054 84 23 . . PUNCT bracis-19054 85 1 a a DET bracis-19054 85 2 new new ADJ bracis-19054 85 3 example example NOUN bracis-19054 85 4 is be AUX bracis-19054 85 5 classified classify VERB bracis-19054 85 6 as as ADP bracis-19054 85 7 the the DET bracis-19054 85 8 interest interest NOUN bracis-19054 85 9 class class NOUN bracis-19054 85 10 if if SCONJ bracis-19054 85 11 its its PRON bracis-19054 85 12 distance distance NOUN bracis-19054 85 13 to to ADP bracis-19054 85 14 the the DET bracis-19054 85 15 class class NOUN bracis-19054 85 16 vector vector NOUN bracis-19054 85 17 is be AUX bracis-19054 85 18 below below ADP bracis-19054 85 19 a a DET bracis-19054 85 20 threshold threshold NOUN bracis-19054 85 21 . . PUNCT bracis-19054 86 1 the the DET bracis-19054 86 2 algorithm algorithm NOUN bracis-19054 86 3 is be AUX bracis-19054 86 4 executed execute VERB bracis-19054 86 5 using use VERB bracis-19054 86 6 90 90 NUM bracis-19054 86 7 % % NOUN bracis-19054 86 8 fake fake ADJ bracis-19054 86 9 news news NOUN bracis-19054 86 10 to to PART bracis-19054 86 11 train train VERB bracis-19054 86 12 the the DET bracis-19054 86 13 model model NOUN bracis-19054 86 14 . . PUNCT bracis-19054 87 1 the the DET bracis-19054 87 2 approach approach NOUN bracis-19054 87 3 reaches reach VERB bracis-19054 87 4 an an DET bracis-19054 87 5 average average ADJ bracis-19054 87 6 \(f_1\ \(f_1\ NOUN bracis-19054 87 7 ) ) PUNCT bracis-19054 87 8 that that PRON bracis-19054 87 9 ranges range VERB bracis-19054 87 10 from from ADP bracis-19054 87 11 54 54 NUM bracis-19054 87 12 % % NOUN bracis-19054 87 13 to to PART bracis-19054 87 14 67 67 NUM bracis-19054 87 15 % % NOUN bracis-19054 87 16 , , PUNCT bracis-19054 87 17 especially especially ADV bracis-19054 87 18 considering consider VERB bracis-19054 87 19 the the DET bracis-19054 87 20 fake.br fake.br PROPN bracis-19054 87 21 dataset dataset NOUN bracis-19054 87 22 . . PUNCT bracis-19054 88 1 it it PRON bracis-19054 88 2 is be AUX bracis-19054 88 3 possible possible ADJ bracis-19054 88 4 to to PART bracis-19054 88 5 observe observe VERB bracis-19054 88 6 that that SCONJ bracis-19054 88 7 most most ADJ bracis-19054 88 8 of of ADP bracis-19054 88 9 the the DET bracis-19054 88 10 existing exist VERB bracis-19054 88 11 work work NOUN bracis-19054 88 12 for for ADP bracis-19054 88 13 fake fake ADJ bracis-19054 88 14 news news NOUN bracis-19054 88 15 detection detection NOUN bracis-19054 88 16 adopts adopt VERB bracis-19054 88 17 binary binary ADJ bracis-19054 88 18 supervised supervised ADJ bracis-19054 88 19 learning learning NOUN bracis-19054 88 20 approaches approach NOUN bracis-19054 88 21 , , PUNCT bracis-19054 88 22 which which PRON bracis-19054 88 23 requires require VERB bracis-19054 88 24 labeling labeling NOUN bracis-19054 88 25 news news NOUN bracis-19054 88 26 considering consider VERB bracis-19054 88 27 the the DET bracis-19054 88 28 real real ADJ bracis-19054 88 29 and and CCONJ bracis-19054 88 30 fake fake ADJ bracis-19054 88 31 classes class NOUN bracis-19054 88 32 . . PUNCT bracis-19054 89 1 additionally additionally ADV bracis-19054 89 2 , , PUNCT bracis-19054 89 3 the the DET bracis-19054 89 4 broad broad ADJ bracis-19054 89 5 spectrum spectrum NOUN bracis-19054 89 6 of of ADP bracis-19054 89 7 real real ADJ bracis-19054 89 8 news news NOUN bracis-19054 89 9 also also ADV bracis-19054 89 10 makes make VERB bracis-19054 89 11 the the DET bracis-19054 89 12 generation generation NOUN bracis-19054 89 13 of of ADP bracis-19054 89 14 a a DET bracis-19054 89 15 representative representative ADJ bracis-19054 89 16 labeled label VERB bracis-19054 89 17 training training NOUN bracis-19054 89 18 set set VERB bracis-19054 89 19 difficult difficult ADJ bracis-19054 89 20 . . PUNCT bracis-19054 90 1 on on ADP bracis-19054 90 2 the the DET bracis-19054 90 3 other other ADJ bracis-19054 90 4 hand hand NOUN bracis-19054 90 5 , , PUNCT bracis-19054 90 6 the the DET bracis-19054 90 7 pul pul NOUN bracis-19054 90 8 algorithms algorithm NOUN bracis-19054 90 9 , , PUNCT bracis-19054 90 10 which which PRON bracis-19054 90 11 mitigate mitigate VERB bracis-19054 90 12 the the DET bracis-19054 90 13 drawback drawback NOUN bracis-19054 90 14 of of ADP bracis-19054 90 15 binary binary NOUN bracis-19054 90 16 supervised supervised ADJ bracis-19054 90 17 learning learn VERB bracis-19054 90 18 algorithms algorithm NOUN bracis-19054 90 19 in in ADP bracis-19054 90 20 fake fake ADJ bracis-19054 90 21 news news NOUN bracis-19054 90 22 detection detection NOUN bracis-19054 90 23 , , PUNCT bracis-19054 90 24 are be AUX bracis-19054 90 25 scarce scarce ADJ bracis-19054 90 26 in in ADP bracis-19054 90 27 this this DET bracis-19054 90 28 scenario scenario NOUN bracis-19054 90 29 . . PUNCT bracis-19054 91 1 also also ADV bracis-19054 91 2 , , PUNCT bracis-19054 91 3 the the DET bracis-19054 91 4 use use NOUN bracis-19054 91 5 of of ADP bracis-19054 91 6 heterogeneous heterogeneous ADJ bracis-19054 91 7 networks network NOUN bracis-19054 91 8 , , PUNCT bracis-19054 91 9 which which PRON bracis-19054 91 10 usually usually ADV bracis-19054 91 11 are be AUX bracis-19054 91 12 more more ADV bracis-19054 91 13 adequate adequate ADJ bracis-19054 91 14 in in ADP bracis-19054 91 15 semi semi ADJ bracis-19054 91 16 - - ADJ bracis-19054 91 17 supervised supervised ADJ bracis-19054 91 18 text text NOUN bracis-19054 91 19 classification classification NOUN bracis-19054 91 20 [ [ X bracis-19054 91 21 12 12 NUM bracis-19054 91 22 ] ] PUNCT bracis-19054 91 23 , , PUNCT bracis-19054 91 24 has have AUX bracis-19054 91 25 not not PART bracis-19054 91 26 been be AUX bracis-19054 91 27 explored explore VERB bracis-19054 91 28 in in ADP bracis-19054 91 29 the the DET bracis-19054 91 30 literature literature NOUN bracis-19054 91 31 for for ADP bracis-19054 91 32 detecting detect VERB bracis-19054 91 33 fake fake ADJ bracis-19054 91 34 news news NOUN bracis-19054 91 35 . . PUNCT bracis-19054 92 1 considering consider VERB bracis-19054 92 2 these these DET bracis-19054 92 3 gaps gap NOUN bracis-19054 92 4 , , PUNCT bracis-19054 92 5 in in ADP bracis-19054 92 6 this this DET bracis-19054 92 7 paper paper NOUN bracis-19054 92 8 we we PRON bracis-19054 92 9 propose propose VERB bracis-19054 92 10 a a DET bracis-19054 92 11 pul pul NOUN bracis-19054 92 12 algorithm algorithm NOUN bracis-19054 92 13 for for ADP bracis-19054 92 14 fake fake ADJ bracis-19054 92 15 - - PUNCT bracis-19054 92 16 news news NOUN bracis-19054 92 17 detection detection NOUN bracis-19054 92 18 , , PUNCT bracis-19054 92 19 which which PRON bracis-19054 92 20 is be AUX bracis-19054 92 21 based base VERB bracis-19054 92 22 on on ADP bracis-19054 92 23 a a DET bracis-19054 92 24 heterogeneous heterogeneous ADJ bracis-19054 92 25 network network NOUN bracis-19054 92 26 composed compose VERB bracis-19054 92 27 of of ADP bracis-19054 92 28 news news NOUN bracis-19054 92 29 , , PUNCT bracis-19054 92 30 terms term NOUN bracis-19054 92 31 , , PUNCT bracis-19054 92 32 and and CCONJ bracis-19054 92 33 linguistic linguistic ADJ bracis-19054 92 34 features feature NOUN bracis-19054 92 35 . . PUNCT bracis-19054 93 1 our our PRON bracis-19054 93 2 approach approach NOUN bracis-19054 93 3 uses use VERB bracis-19054 93 4 only only ADJ bracis-19054 93 5 information information NOUN bracis-19054 93 6 that that PRON bracis-19054 93 7 can can AUX bracis-19054 93 8 be be AUX bracis-19054 93 9 collected collect VERB bracis-19054 93 10 in in ADP bracis-19054 93 11 the the DET bracis-19054 93 12 publication publication NOUN bracis-19054 93 13 ’s ’s PART bracis-19054 93 14 content content NOUN bracis-19054 93 15 , , PUNCT bracis-19054 93 16 not not PART bracis-19054 93 17 depending depend VERB bracis-19054 93 18 on on ADP bracis-19054 93 19 external external ADJ bracis-19054 93 20 information information NOUN bracis-19054 93 21 . . PUNCT bracis-19054 94 1 the the DET bracis-19054 94 2 proposed propose VERB bracis-19054 94 3 approach approach NOUN bracis-19054 94 4 is be AUX bracis-19054 94 5 detailed detail VERB bracis-19054 94 6 in in ADP bracis-19054 94 7 the the DET bracis-19054 94 8 next next ADJ bracis-19054 94 9 sections section NOUN bracis-19054 94 10 . . PUNCT bracis-19054 95 1 3 3 NUM bracis-19054 95 2 positive positive ADJ bracis-19054 95 3 and and CCONJ bracis-19054 95 4     SPACE bracis-19054 95 5 unlabeled unlabele VERB bracis-19054 95 6 learning learn VERB bracis-19054 95 7 algorithms algorithm NOUN bracis-19054 95 8 pul pul NOUN bracis-19054 95 9 algorithms algorithm NOUN bracis-19054 95 10 learn learn VERB bracis-19054 95 11 models model NOUN bracis-19054 95 12 considering consider VERB bracis-19054 95 13 a a DET bracis-19054 95 14 set set NOUN bracis-19054 95 15 of of ADP bracis-19054 95 16 labeled label VERB bracis-19054 95 17 documents document NOUN bracis-19054 95 18 of of ADP bracis-19054 95 19 an an DET bracis-19054 95 20 interest interest NOUN bracis-19054 95 21 class class NOUN bracis-19054 95 22 and and CCONJ bracis-19054 95 23 unlabeled unlabeled ADJ bracis-19054 95 24 documents document NOUN bracis-19054 95 25 to to PART bracis-19054 95 26 train train VERB bracis-19054 95 27 a a DET bracis-19054 95 28 classifier classifier NOUN bracis-19054 95 29 ( ( PUNCT bracis-19054 95 30 inductive inductive ADJ bracis-19054 95 31 semi semi ADJ bracis-19054 95 32 - - ADJ bracis-19054 95 33 supervised supervised ADJ bracis-19054 95 34 learning learning NOUN bracis-19054 95 35 ) ) PUNCT bracis-19054 95 36 or or CCONJ bracis-19054 95 37 to to PART bracis-19054 95 38 classify classify VERB bracis-19054 95 39 unlabeled unlabele VERB bracis-19054 95 40 known know VERB bracis-19054 95 41 documents document NOUN bracis-19054 95 42 ( ( PUNCT bracis-19054 95 43 transductive transductive ADJ bracis-19054 95 44 semi semi ADJ bracis-19054 95 45 - - ADJ bracis-19054 95 46 supervised supervised ADJ bracis-19054 95 47 learning learning NOUN bracis-19054 95 48 ) ) PUNCT bracis-19054 96 1 [ [ X bracis-19054 96 2 3 3 NUM bracis-19054 96 3 ] ] PUNCT bracis-19054 96 4 . . PUNCT bracis-19054 97 1 the the DET bracis-19054 97 2 purpose purpose NOUN bracis-19054 97 3 of of ADP bracis-19054 97 4 using use VERB bracis-19054 97 5 unlabeled unlabeled ADJ bracis-19054 97 6 documents document NOUN bracis-19054 97 7 during during ADP bracis-19054 97 8 learning learn VERB bracis-19054 97 9 is be AUX bracis-19054 97 10 to to PART bracis-19054 97 11 improve improve VERB bracis-19054 97 12 classification classification NOUN bracis-19054 97 13 performance performance NOUN bracis-19054 97 14 . . PUNCT bracis-19054 98 1 besides besides SCONJ bracis-19054 98 2 , , PUNCT bracis-19054 98 3 since since SCONJ bracis-19054 98 4 the the DET bracis-19054 98 5 unlabeled unlabeled ADJ bracis-19054 98 6 documents document NOUN bracis-19054 98 7 are be AUX bracis-19054 98 8 easy easy ADJ bracis-19054 98 9 to to PART bracis-19054 98 10 collect collect VERB bracis-19054 98 11 and and CCONJ bracis-19054 98 12 a a DET bracis-19054 98 13 user user NOUN bracis-19054 98 14 has have VERB bracis-19054 98 15 to to PART bracis-19054 98 16 label label VERB bracis-19054 98 17 only only ADV bracis-19054 98 18 a a DET bracis-19054 98 19 few few ADJ bracis-19054 98 20 documents document NOUN bracis-19054 98 21 of of ADP bracis-19054 98 22 the the DET bracis-19054 98 23 interest interest NOUN bracis-19054 98 24 class class NOUN bracis-19054 98 25 , , PUNCT bracis-19054 98 26 pul pul NOUN bracis-19054 98 27 has have AUX bracis-19054 98 28 gained gain VERB bracis-19054 98 29 attention attention NOUN bracis-19054 98 30 in in ADP bracis-19054 98 31 the the DET bracis-19054 98 32 last last ADJ bracis-19054 98 33 years year NOUN bracis-19054 99 1 [ [ X bracis-19054 99 2 3 3 NUM bracis-19054 99 3 , , PUNCT bracis-19054 99 4 9 9 NUM bracis-19054 99 5 ] ] PUNCT bracis-19054 99 6 . . PUNCT bracis-19054 100 1 the the DET bracis-19054 100 2 most most ADV bracis-19054 100 3 common common ADJ bracis-19054 100 4 pul pul NOUN bracis-19054 100 5 approaches approach NOUN bracis-19054 100 6 are be AUX bracis-19054 100 7 those those PRON bracis-19054 100 8 which which PRON bracis-19054 100 9 perform perform VERB bracis-19054 100 10 learning learn VERB bracis-19054 100 11 in in ADP bracis-19054 100 12 two two NUM bracis-19054 100 13 stages stage NOUN bracis-19054 100 14 . . PUNCT bracis-19054 101 1 in in ADP bracis-19054 101 2 the the DET bracis-19054 101 3 first first ADJ bracis-19054 101 4 stage stage NOUN bracis-19054 101 5 , , PUNCT bracis-19054 101 6 a a DET bracis-19054 101 7 set set NOUN bracis-19054 101 8 of of ADP bracis-19054 101 9 non non ADJ bracis-19054 101 10 - - ADJ bracis-19054 101 11 interest interest ADJ bracis-19054 101 12 documents document NOUN bracis-19054 101 13 is be AUX bracis-19054 101 14 generated generate VERB bracis-19054 101 15 by by ADP bracis-19054 101 16 extracting extract VERB bracis-19054 101 17 reliable reliable ADJ bracis-19054 101 18 outliers outlier NOUN bracis-19054 101 19 documents document NOUN bracis-19054 101 20 to to PART bracis-19054 101 21 be be AUX bracis-19054 101 22 considered consider VERB bracis-19054 101 23 as as ADP bracis-19054 101 24 non non ADJ bracis-19054 101 25 - - ADJ bracis-19054 101 26 interest interest ADJ bracis-19054 101 27 class class NOUN bracis-19054 101 28 . . PUNCT bracis-19054 102 1 in in ADP bracis-19054 102 2 addition addition NOUN bracis-19054 102 3 , , PUNCT bracis-19054 102 4 the the DET bracis-19054 102 5 set set NOUN bracis-19054 102 6 of of ADP bracis-19054 102 7 interest interest NOUN bracis-19054 102 8 documents document NOUN bracis-19054 102 9 can can AUX bracis-19054 102 10 also also ADV bracis-19054 102 11 be be AUX bracis-19054 102 12 increased increase VERB bracis-19054 102 13 with with ADP bracis-19054 102 14 reliable reliable ADJ bracis-19054 102 15 interest interest NOUN bracis-19054 102 16 documents document NOUN bracis-19054 102 17 . . PUNCT bracis-19054 103 1 once once ADV bracis-19054 103 2 there there PRON bracis-19054 103 3 are be VERB bracis-19054 103 4 interest interest NOUN bracis-19054 103 5 , , PUNCT bracis-19054 103 6 non non ADJ bracis-19054 103 7 - - NOUN bracis-19054 103 8 interest interest NOUN bracis-19054 103 9 , , PUNCT bracis-19054 103 10 and and CCONJ bracis-19054 103 11 unlabeled unlabeled ADJ bracis-19054 103 12 documents document NOUN bracis-19054 103 13 , , PUNCT bracis-19054 103 14 a a DET bracis-19054 103 15 transductive transductive ADJ bracis-19054 103 16 learning learn VERB bracis-19054 103 17 algorithm algorithm NOUN bracis-19054 103 18 is be AUX bracis-19054 103 19 applied apply VERB bracis-19054 103 20 to to PART bracis-19054 103 21 infer infer VERB bracis-19054 103 22 the the DET bracis-19054 103 23 label label NOUN bracis-19054 103 24 to to ADP bracis-19054 103 25 the the DET bracis-19054 103 26 remaining remain VERB bracis-19054 103 27 unlabeled unlabele VERB bracis-19054 103 28 documents document NOUN bracis-19054 103 29 in in ADP bracis-19054 103 30 the the DET bracis-19054 103 31 second second ADJ bracis-19054 103 32 stage stage NOUN bracis-19054 103 33 [ [ X bracis-19054 103 34 3 3 NUM bracis-19054 103 35 ] ] PUNCT bracis-19054 103 36 . . PUNCT bracis-19054 104 1 generally generally ADV bracis-19054 104 2 , , PUNCT bracis-19054 104 3 the the DET bracis-19054 104 4 pul pul NOUN bracis-19054 104 5 algorithms algorithm NOUN bracis-19054 104 6 based base VERB bracis-19054 104 7 on on ADP bracis-19054 104 8 this this DET bracis-19054 104 9 framework framework NOUN bracis-19054 104 10 include include VERB bracis-19054 104 11 the the DET bracis-19054 104 12 application application NOUN bracis-19054 104 13 of of ADP bracis-19054 104 14 self self NOUN bracis-19054 104 15 - - PUNCT bracis-19054 104 16 training training NOUN bracis-19054 104 17 in in ADP bracis-19054 104 18 some some PRON bracis-19054 104 19 or or CCONJ bracis-19054 104 20 all all PRON bracis-19054 104 21 of of ADP bracis-19054 104 22 the the DET bracis-19054 104 23 steps step NOUN bracis-19054 104 24 , , PUNCT bracis-19054 104 25 which which PRON bracis-19054 104 26 can can AUX bracis-19054 104 27 be be AUX bracis-19054 104 28 computationally computationally ADV bracis-19054 104 29 costly costly ADJ bracis-19054 104 30 and and CCONJ bracis-19054 104 31 do do AUX bracis-19054 104 32 not not PART bracis-19054 104 33 improve improve VERB bracis-19054 104 34 the the DET bracis-19054 104 35 classification classification NOUN bracis-19054 104 36 performance performance NOUN bracis-19054 104 37 [ [ X bracis-19054 104 38 12 12 NUM bracis-19054 104 39 ] ] PUNCT bracis-19054 104 40 . . PUNCT bracis-19054 105 1 we we PRON bracis-19054 105 2 can can AUX bracis-19054 105 3 observe observe VERB bracis-19054 105 4 the the DET bracis-19054 105 5 use use NOUN bracis-19054 105 6 of of ADP bracis-19054 105 7 algorithms algorithm NOUN bracis-19054 105 8 based base VERB bracis-19054 105 9 on on ADP bracis-19054 105 10 vector vector NOUN bracis-19054 105 11 - - PUNCT bracis-19054 105 12 space space NOUN bracis-19054 105 13 model model NOUN bracis-19054 105 14 , , PUNCT bracis-19054 105 15 such such ADJ bracis-19054 105 16 as as ADP bracis-19054 105 17 rocchio rocchio NOUN bracis-19054 105 18 or or CCONJ bracis-19054 105 19 expectation expectation NOUN bracis-19054 105 20 maximization maximization NOUN bracis-19054 105 21 , , PUNCT bracis-19054 105 22 also also ADV bracis-19054 105 23 demonstrated demonstrate VERB bracis-19054 105 24 lower low ADJ bracis-19054 105 25 performances performance NOUN bracis-19054 105 26 than than ADP bracis-19054 105 27 other other ADJ bracis-19054 105 28 approaches approach NOUN bracis-19054 105 29 in in ADP bracis-19054 105 30 text text NOUN bracis-19054 105 31 classification classification NOUN bracis-19054 106 1 [ [ X bracis-19054 106 2 12 12 NUM bracis-19054 106 3 , , PUNCT bracis-19054 106 4 20 20 NUM bracis-19054 106 5 ] ] PUNCT bracis-19054 106 6 . . PUNCT bracis-19054 107 1 despite despite SCONJ bracis-19054 107 2 the the DET bracis-19054 107 3 benefits benefit NOUN bracis-19054 107 4 of of ADP bracis-19054 107 5 network network NOUN bracis-19054 107 6 - - PUNCT bracis-19054 107 7 based base VERB bracis-19054 107 8 approaches approach NOUN bracis-19054 107 9 in in ADP bracis-19054 107 10 semi semi ADJ bracis-19054 107 11 - - ADJ bracis-19054 107 12 supervised supervised ADJ bracis-19054 107 13 learning learning NOUN bracis-19054 107 14 scenarios scenario NOUN bracis-19054 107 15 , , PUNCT bracis-19054 107 16 there there PRON bracis-19054 107 17 are be VERB bracis-19054 107 18 few few ADJ bracis-19054 107 19 pul pul NOUN bracis-19054 107 20 proposals proposal NOUN bracis-19054 107 21 based base VERB bracis-19054 107 22 on on ADP bracis-19054 107 23 networks network NOUN bracis-19054 107 24 , , PUNCT bracis-19054 107 25 such such ADJ bracis-19054 107 26 as as ADP bracis-19054 107 27 positive positive ADJ bracis-19054 107 28 documents document NOUN bracis-19054 107 29 enlarging enlarge VERB bracis-19054 107 30 pu pu PROPN bracis-19054 107 31 classifier classifier NOUN bracis-19054 107 32 ( ( PUNCT bracis-19054 107 33 pe pe NOUN bracis-19054 107 34 - - PUNCT bracis-19054 107 35 puc puc NOUN bracis-19054 107 36 ) ) PUNCT bracis-19054 108 1 [ [ X bracis-19054 108 2 20 20 NUM bracis-19054 108 3 ] ] PUNCT bracis-19054 108 4 and and CCONJ bracis-19054 108 5 positive positive ADJ bracis-19054 108 6 and and CCONJ bracis-19054 108 7 unlabeled unlabeled ADJ bracis-19054 108 8 learning learning NOUN bracis-19054 108 9 by by ADP bracis-19054 108 10 label label NOUN bracis-19054 108 11 propagation propagation NOUN bracis-19054 108 12 ( ( PUNCT bracis-19054 108 13 pu pu NOUN bracis-19054 108 14 - - PUNCT bracis-19054 108 15 lp lp NOUN bracis-19054 108 16 ) ) PUNCT bracis-19054 109 1 [ [ X bracis-19054 109 2 9 9 NUM bracis-19054 109 3 ] ] PUNCT bracis-19054 109 4 . . PUNCT bracis-19054 110 1 pe pe ADJ bracis-19054 110 2 - - PUNCT bracis-19054 110 3 puc puc NOUN bracis-19054 110 4 uses use VERB bracis-19054 110 5 network network NOUN bracis-19054 110 6 representations representation NOUN bracis-19054 110 7 only only ADV bracis-19054 110 8 to to PART bracis-19054 110 9 increase increase VERB bracis-19054 110 10 the the DET bracis-19054 110 11 set set NOUN bracis-19054 110 12 of of ADP bracis-19054 110 13 interest interest NOUN bracis-19054 110 14 documents document NOUN bracis-19054 110 15 with with ADP bracis-19054 110 16 reliable reliable ADJ bracis-19054 110 17 positive positive ADJ bracis-19054 110 18 documents document NOUN bracis-19054 110 19 . . PUNCT bracis-19054 111 1 the the DET bracis-19054 111 2 network network NOUN bracis-19054 111 3 is be AUX bracis-19054 111 4 not not PART bracis-19054 111 5 used use VERB bracis-19054 111 6 in in ADP bracis-19054 111 7 the the DET bracis-19054 111 8 classification classification NOUN bracis-19054 111 9 step step NOUN bracis-19054 111 10 . . PUNCT bracis-19054 112 1 the the DET bracis-19054 112 2 remaining remain VERB bracis-19054 112 3 steps step NOUN bracis-19054 112 4 consider consider VERB bracis-19054 112 5 bayesian bayesian NOUN bracis-19054 112 6 classifiers classifier NOUN bracis-19054 112 7 , , PUNCT bracis-19054 112 8 which which PRON bracis-19054 112 9 have have AUX bracis-19054 112 10 been be AUX bracis-19054 112 11 shown show VERB bracis-19054 112 12 to to PART bracis-19054 112 13 perform perform VERB bracis-19054 112 14 poorly poorly ADV bracis-19054 112 15 when when SCONJ bracis-19054 112 16 the the DET bracis-19054 112 17 labeled label VERB bracis-19054 112 18 training training NOUN bracis-19054 112 19 data datum NOUN bracis-19054 112 20 set set VERB bracis-19054 112 21 is be AUX bracis-19054 112 22 small small ADJ bracis-19054 112 23 [ [ X bracis-19054 112 24 12 12 NUM bracis-19054 112 25 ] ] PUNCT bracis-19054 112 26 . . PUNCT bracis-19054 113 1 the the DET bracis-19054 113 2 use use NOUN bracis-19054 113 3 of of ADP bracis-19054 113 4 networks network NOUN bracis-19054 113 5 allows allow VERB bracis-19054 113 6 to to PART bracis-19054 113 7 extract extract VERB bracis-19054 113 8 patterns pattern NOUN bracis-19054 113 9 of of ADP bracis-19054 113 10 the the DET bracis-19054 113 11 interest interest NOUN bracis-19054 113 12 class class NOUN bracis-19054 113 13 even even ADV bracis-19054 113 14 they they PRON bracis-19054 113 15 are be AUX bracis-19054 113 16 given give VERB bracis-19054 113 17 by by ADP bracis-19054 113 18 different different ADJ bracis-19054 113 19 distributions distribution NOUN bracis-19054 113 20 , , PUNCT bracis-19054 113 21 densities density NOUN bracis-19054 113 22 on on ADP bracis-19054 113 23 regions region NOUN bracis-19054 113 24 in in ADP bracis-19054 113 25 the the DET bracis-19054 113 26 space space NOUN bracis-19054 113 27 [ [ X bracis-19054 113 28 15 15 NUM bracis-19054 113 29 ] ] PUNCT bracis-19054 113 30 . . PUNCT bracis-19054 114 1 as as SCONJ bracis-19054 114 2 pu pu PROPN bracis-19054 114 3 - - PUNCT bracis-19054 114 4 lp lp PROPN bracis-19054 114 5 is be AUX bracis-19054 114 6 entirely entirely ADV bracis-19054 114 7 network network NOUN bracis-19054 114 8 - - PUNCT bracis-19054 114 9 based base VERB bracis-19054 114 10 , , PUNCT bracis-19054 114 11 achieving achieve VERB bracis-19054 114 12 good good ADJ bracis-19054 114 13 performance performance NOUN bracis-19054 114 14 in in ADP bracis-19054 114 15 [ [ X bracis-19054 114 16 9 9 NUM bracis-19054 114 17 ] ] PUNCT bracis-19054 114 18 , , PUNCT bracis-19054 114 19 we we PRON bracis-19054 114 20 propose propose VERB bracis-19054 114 21 to to PART bracis-19054 114 22 use use VERB bracis-19054 114 23 it it PRON bracis-19054 114 24 for for ADP bracis-19054 114 25 news news NOUN bracis-19054 114 26 classification classification NOUN bracis-19054 114 27 . . PUNCT bracis-19054 115 1 next next ADJ bracis-19054 115 2 section section NOUN bracis-19054 115 3 describes describe VERB bracis-19054 115 4 the the DET bracis-19054 115 5 proposed propose VERB bracis-19054 115 6 approach approach NOUN bracis-19054 115 7 . . PUNCT bracis-19054 116 1 4 4 NUM bracis-19054 116 2 proposed propose VERB bracis-19054 116 3 approach approach NOUN bracis-19054 116 4 : : PUNCT bracis-19054 116 5 pu pu PROPN bracis-19054 116 6 - - PUNCT bracis-19054 116 7 lp lp NOUN bracis-19054 116 8 for for ADP bracis-19054 116 9     SPACE bracis-19054 116 10 fake fake ADJ bracis-19054 116 11 news news NOUN bracis-19054 116 12 detection detection NOUN bracis-19054 116 13 considering consider VERB bracis-19054 116 14 that that SCONJ bracis-19054 116 15 pul pul NOUN bracis-19054 116 16 algorithms algorithm NOUN bracis-19054 116 17 can can AUX bracis-19054 116 18 reduce reduce VERB bracis-19054 116 19 labeling labeling NOUN bracis-19054 116 20 effort effort NOUN bracis-19054 116 21 for for ADP bracis-19054 116 22 news news NOUN bracis-19054 116 23 classification classification NOUN bracis-19054 116 24 , , PUNCT bracis-19054 116 25 the the DET bracis-19054 116 26 benefits benefit NOUN bracis-19054 116 27 of of ADP bracis-19054 116 28 network network NOUN bracis-19054 116 29 - - PUNCT bracis-19054 116 30 based base VERB bracis-19054 116 31 representations representation NOUN bracis-19054 116 32 for for ADP bracis-19054 116 33 semi semi ADJ bracis-19054 116 34 - - ADJ bracis-19054 116 35 supervised supervised ADJ bracis-19054 116 36 learning learning NOUN bracis-19054 116 37 , , PUNCT bracis-19054 116 38 and and CCONJ bracis-19054 116 39 that that SCONJ bracis-19054 116 40 adding add VERB bracis-19054 116 41 extra extra ADJ bracis-19054 116 42 information information NOUN bracis-19054 116 43 to to ADP bracis-19054 116 44 the the DET bracis-19054 116 45 network network NOUN bracis-19054 116 46 can can AUX bracis-19054 116 47 contribute contribute VERB bracis-19054 116 48 to to ADP bracis-19054 116 49 differentiating differentiate VERB bracis-19054 116 50 real real ADJ bracis-19054 116 51 and and CCONJ bracis-19054 116 52 fake fake ADJ bracis-19054 116 53 content content NOUN bracis-19054 116 54 , , PUNCT bracis-19054 116 55 we we PRON bracis-19054 116 56 propose propose VERB bracis-19054 116 57 an an DET bracis-19054 116 58 approach approach NOUN bracis-19054 116 59 based base VERB bracis-19054 116 60 on on ADP bracis-19054 116 61 pu pu PROPN bracis-19054 116 62 - - PUNCT bracis-19054 116 63 lp lp PROPN bracis-19054 116 64 algorithm algorithm NOUN bracis-19054 116 65 applied apply VERB bracis-19054 116 66 to to ADP bracis-19054 116 67 fake fake ADJ bracis-19054 116 68 news news NOUN bracis-19054 116 69 detection detection NOUN bracis-19054 116 70 . . PUNCT bracis-19054 117 1 after after SCONJ bracis-19054 117 2 pu pu PROPN bracis-19054 117 3 - - PUNCT bracis-19054 117 4 lp lp PROPN bracis-19054 117 5 infer infer NOUN bracis-19054 117 6 interest interest NOUN bracis-19054 117 7 and and CCONJ bracis-19054 117 8 non non ADJ bracis-19054 117 9 - - ADJ bracis-19054 117 10 interest interest ADJ bracis-19054 117 11 news news NOUN bracis-19054 117 12 sets set NOUN bracis-19054 117 13 in in ADP bracis-19054 117 14 the the DET bracis-19054 117 15 news news NOUN bracis-19054 117 16 network network NOUN bracis-19054 117 17 , , PUNCT bracis-19054 117 18 we we PRON bracis-19054 117 19 add add VERB bracis-19054 117 20 relations relation NOUN bracis-19054 117 21 ( ( PUNCT bracis-19054 117 22 edges edge NOUN bracis-19054 117 23 ) ) PUNCT bracis-19054 117 24 between between ADP bracis-19054 117 25 news news NOUN bracis-19054 117 26 , , PUNCT bracis-19054 117 27 relevant relevant ADJ bracis-19054 117 28 terms term NOUN bracis-19054 117 29 and and CCONJ bracis-19054 117 30 linguistic linguistic ADJ bracis-19054 117 31 features feature NOUN bracis-19054 117 32 making make VERB bracis-19054 117 33 a a DET bracis-19054 117 34 heterogeneous heterogeneous ADJ bracis-19054 117 35 network network NOUN bracis-19054 117 36 . . PUNCT bracis-19054 118 1 thus thus ADV bracis-19054 118 2 , , PUNCT bracis-19054 118 3 a a DET bracis-19054 118 4 label label NOUN bracis-19054 118 5 propagation propagation NOUN bracis-19054 118 6 is be AUX bracis-19054 118 7 applied apply VERB bracis-19054 118 8 to to PART bracis-19054 118 9 classify classify VERB bracis-19054 118 10 the the DET bracis-19054 118 11 unlabeled unlabeled ADJ bracis-19054 118 12 news news NOUN bracis-19054 118 13 . . PUNCT bracis-19054 119 1 we we PRON bracis-19054 119 2 compared compare VERB bracis-19054 119 3 different different ADJ bracis-19054 119 4 proposed propose VERB bracis-19054 119 5 heterogeneous heterogeneous ADJ bracis-19054 119 6 networks network NOUN bracis-19054 119 7 with with ADP bracis-19054 119 8 the the DET bracis-19054 119 9 traditional traditional ADJ bracis-19054 119 10 pu pu PROPN bracis-19054 119 11 - - PUNCT bracis-19054 119 12 lp lp PROPN bracis-19054 119 13 algorithm algorithm NOUN bracis-19054 119 14 considering consider VERB bracis-19054 119 15 only only ADV bracis-19054 119 16 the the DET bracis-19054 119 17 news news NOUN bracis-19054 119 18 network network NOUN bracis-19054 119 19 . . PUNCT bracis-19054 120 1 we we PRON bracis-19054 120 2 use use VERB bracis-19054 120 3 different different ADJ bracis-19054 120 4 parameters parameter NOUN bracis-19054 120 5 , , PUNCT bracis-19054 120 6 label label NOUN bracis-19054 120 7 propagation propagation NOUN bracis-19054 120 8 algorithms algorithm NOUN bracis-19054 120 9 , , PUNCT bracis-19054 120 10 and and CCONJ bracis-19054 120 11 representation representation NOUN bracis-19054 120 12 models model NOUN bracis-19054 120 13 for for ADP bracis-19054 120 14 unstructured unstructured ADJ bracis-19054 120 15 data datum NOUN bracis-19054 120 16 to to PART bracis-19054 120 17 assess assess VERB bracis-19054 120 18 the the DET bracis-19054 120 19 algorithms algorithm NOUN bracis-19054 120 20 ’ ' PUNCT bracis-19054 120 21 behavior behavior NOUN bracis-19054 120 22 . . PUNCT bracis-19054 121 1 figure figure VERB bracis-19054 121 2     SPACE bracis-19054 121 3 1 1 NUM bracis-19054 121 4 presents present VERB bracis-19054 121 5 the the DET bracis-19054 121 6 proposed propose VERB bracis-19054 121 7 approach approach NOUN bracis-19054 121 8 of of ADP bracis-19054 121 9 pu pu PROPN bracis-19054 121 10 - - PUNCT bracis-19054 121 11 lp lp PROPN bracis-19054 121 12 for for ADP bracis-19054 121 13 semi semi ADJ bracis-19054 121 14 - - ADJ bracis-19054 121 15 supervised supervised ADJ bracis-19054 121 16 fake fake ADJ bracis-19054 121 17 news news NOUN bracis-19054 121 18 detection detection NOUN bracis-19054 121 19 . . PUNCT bracis-19054 122 1 the the DET bracis-19054 122 2 next next ADJ bracis-19054 122 3 subsections subsection NOUN bracis-19054 122 4 presents present VERB bracis-19054 122 5 details detail NOUN bracis-19054 122 6 of of ADP bracis-19054 122 7 each each DET bracis-19054 122 8 stage stage NOUN bracis-19054 122 9 of of ADP bracis-19054 122 10 the the DET bracis-19054 122 11 proposed propose VERB bracis-19054 122 12 approach approach NOUN bracis-19054 122 13 . . PUNCT bracis-19054 123 1 fig fig NOUN bracis-19054 123 2 . . PUNCT bracis-19054 124 1 1 1 X bracis-19054 124 2 . . NUM bracis-19054 124 3 proposed propose VERB bracis-19054 124 4 approach approach NOUN bracis-19054 124 5 for for ADP bracis-19054 124 6 detecting detect VERB bracis-19054 124 7 fake fake ADJ bracis-19054 124 8 news news NOUN bracis-19054 124 9 based base VERB bracis-19054 124 10 on on ADP bracis-19054 124 11 the the DET bracis-19054 124 12 semi semi ADJ bracis-19054 124 13 - - ADJ bracis-19054 124 14 supervised supervised ADJ bracis-19054 124 15 pu pu PROPN bracis-19054 124 16 - - PUNCT bracis-19054 124 17 lp lp NOUN bracis-19054 124 18 algorithm algorithm NOUN bracis-19054 124 19 . . PUNCT bracis-19054 125 1 circular circular ADJ bracis-19054 125 2 nodes node NOUN bracis-19054 125 3 represent represent VERB bracis-19054 125 4 news news NOUN bracis-19054 125 5 , , PUNCT bracis-19054 125 6 nodes node NOUN bracis-19054 125 7 with with ADP bracis-19054 125 8 the the DET bracis-19054 125 9 letter letter NOUN bracis-19054 125 10 t t PROPN bracis-19054 125 11 are be AUX bracis-19054 125 12 representative representative ADJ bracis-19054 125 13 unigrams unigram NOUN bracis-19054 125 14 and and CCONJ bracis-19054 125 15 bigrams bigram NOUN bracis-19054 125 16 , , PUNCT bracis-19054 125 17 nodes node NOUN bracis-19054 125 18 with with ADP bracis-19054 125 19 the the DET bracis-19054 125 20 letter letter NOUN bracis-19054 125 21 p p PROPN bracis-19054 125 22 correspond correspond NOUN bracis-19054 125 23 to to ADP bracis-19054 125 24 pausality pausality NOUN bracis-19054 125 25 , , PUNCT bracis-19054 125 26 e e NOUN bracis-19054 125 27 emotiveness emotiveness NOUN bracis-19054 125 28 , , PUNCT bracis-19054 125 29 and and CCONJ bracis-19054 125 30 s s VERB bracis-19054 125 31 the the DET bracis-19054 125 32 average average ADJ bracis-19054 125 33 number number NOUN bracis-19054 125 34 of of ADP bracis-19054 125 35 words word NOUN bracis-19054 125 36 per per ADP bracis-19054 125 37 sentence sentence NOUN bracis-19054 125 38 . . PUNCT bracis-19054 126 1 full full ADJ bracis-19054 126 2 size size NOUN bracis-19054 126 3 image image NOUN bracis-19054 126 4 4.1 4.1 NUM bracis-19054 126 5 news news NOUN bracis-19054 126 6 collection collection NOUN bracis-19054 126 7 and and CCONJ bracis-19054 126 8     SPACE bracis-19054 126 9 representation representation NOUN bracis-19054 126 10 model model NOUN bracis-19054 126 11 let let VERB bracis-19054 126 12 \(\mathcal \(\mathcal ADJ bracis-19054 126 13 { { PUNCT bracis-19054 126 14 d d NOUN bracis-19054 126 15 } } PUNCT bracis-19054 126 16 = = SYM bracis-19054 126 17 \{d_1 \{d_1 NOUN bracis-19054 126 18 , , PUNCT bracis-19054 126 19 \ldots \ldots INTJ bracis-19054 126 20 , , PUNCT bracis-19054 126 21 d_l d_l PROPN bracis-19054 126 22 , , PUNCT bracis-19054 126 23 d_{l+1 d_{l+1 PROPN bracis-19054 126 24 } } PUNCT bracis-19054 126 25 , , PUNCT bracis-19054 126 26 \ldots \ldots INTJ bracis-19054 126 27 , , PUNCT bracis-19054 126 28 d_{l+u d_{l+u ADJ bracis-19054 126 29 } } SYM bracis-19054 126 30 \}\ \}\ NUM bracis-19054 126 31 ) ) PUNCT bracis-19054 126 32 be be AUX bracis-19054 126 33 a a DET bracis-19054 126 34 news news NOUN bracis-19054 126 35 set set NOUN bracis-19054 126 36 and and CCONJ bracis-19054 126 37 \(\mathcal \(\mathcal ADJ bracis-19054 126 38 { { PUNCT bracis-19054 126 39 c c NOUN bracis-19054 126 40 } } PUNCT bracis-19054 126 41 = = PROPN bracis-19054 126 42 \big \big PROPN bracis-19054 126 43 \{\textit{interest},\)\ \{\textit{interest},\)\ PROPN bracis-19054 126 44 ( ( PUNCT bracis-19054 126 45 \textit{not \textit{not ADV bracis-19054 126 46 interest}\big interest}\big ADJ bracis-19054 126 47 \}\ \}\ NOUN bracis-19054 126 48 ) ) PUNCT bracis-19054 126 49 , , PUNCT bracis-19054 126 50 i.e. i.e. X bracis-19054 126 51 , , PUNCT bracis-19054 126 52 \(\mathcal \(\mathcal ADJ bracis-19054 126 53 { { PUNCT bracis-19054 126 54 c c NOUN bracis-19054 126 55 } } PUNCT bracis-19054 126 56 = = SYM bracis-19054 126 57 \{\textit{fake \{\textit{fake NOUN bracis-19054 126 58 } } PUNCT bracis-19054 126 59 , , PUNCT bracis-19054 126 60 \textit{real}\}\ \textit{real}\}\ PROPN bracis-19054 126 61 ) ) PUNCT bracis-19054 126 62 , , PUNCT bracis-19054 126 63 be be AUX bracis-19054 126 64 a a DET bracis-19054 126 65 set set NOUN bracis-19054 126 66 of of ADP bracis-19054 126 67 class class NOUN bracis-19054 126 68 labels label NOUN bracis-19054 126 69 . . PUNCT bracis-19054 127 1 the the DET bracis-19054 127 2 first first ADJ bracis-19054 127 3 l l NOUN bracis-19054 127 4 elements element NOUN bracis-19054 127 5 of of ADP bracis-19054 127 6 \(\mathcal \(\mathcal ADJ bracis-19054 127 7 { { PUNCT bracis-19054 127 8 d}\ d}\ NOUN bracis-19054 127 9 ) ) PUNCT bracis-19054 127 10 are be AUX bracis-19054 127 11 fake fake ADJ bracis-19054 127 12 labeled label VERB bracis-19054 127 13 news news NOUN bracis-19054 127 14 , , PUNCT bracis-19054 127 15 composing compose VERB bracis-19054 127 16 the the DET bracis-19054 127 17 interest interest NOUN bracis-19054 127 18 labeled label VERB bracis-19054 127 19 set set VERB bracis-19054 127 20 \(\mathcal \(\mathcal ADJ bracis-19054 127 21 { { PUNCT bracis-19054 127 22 d}^{+}\ d}^{+}\ NOUN bracis-19054 127 23 ) ) PUNCT bracis-19054 127 24 . . PUNCT bracis-19054 128 1 the the DET bracis-19054 128 2 remaining remain VERB bracis-19054 128 3 u u NOUN bracis-19054 128 4 elements element NOUN bracis-19054 128 5 are be AUX bracis-19054 128 6 unlabeled unlabele VERB bracis-19054 128 7 news news NOUN bracis-19054 128 8 ( ( PUNCT bracis-19054 128 9 fake fake ADJ bracis-19054 128 10 and and CCONJ bracis-19054 128 11 real real ADJ bracis-19054 128 12 ) ) PUNCT bracis-19054 128 13 , , PUNCT bracis-19054 128 14 composing compose VERB bracis-19054 128 15 the the DET bracis-19054 128 16 set set NOUN bracis-19054 128 17 \(\mathcal \(\mathcal ADJ bracis-19054 128 18 { { PUNCT bracis-19054 128 19 d}^{u}\ d}^{u}\ NOUN bracis-19054 128 20 ) ) PUNCT bracis-19054 128 21 , , PUNCT bracis-19054 128 22 and and CCONJ bracis-19054 128 23 \(u \(u NOUN bracis-19054 128 24 \gg \gg NOUN bracis-19054 128 25 l\ l\ PROPN bracis-19054 128 26 ) ) PUNCT bracis-19054 128 27 ( ( PUNCT bracis-19054 128 28 fig fig NOUN bracis-19054 128 29 . . PUNCT bracis-19054 128 30     SPACE bracis-19054 128 31 1 1 NUM bracis-19054 128 32 stage stage NOUN bracis-19054 128 33 1 1 NUM bracis-19054 128 34 ) ) PUNCT bracis-19054 128 35 . . PUNCT bracis-19054 129 1 the the DET bracis-19054 129 2 news news NOUN bracis-19054 129 3 dataset dataset NOUN bracis-19054 129 4 must must AUX bracis-19054 129 5 be be AUX bracis-19054 129 6 pre pre VERB bracis-19054 129 7 - - VERB bracis-19054 129 8 processed process VERB bracis-19054 129 9 , , PUNCT bracis-19054 129 10 and and CCONJ bracis-19054 129 11 a a DET bracis-19054 129 12 representation representation NOUN bracis-19054 129 13 model model NOUN bracis-19054 129 14 , , PUNCT bracis-19054 129 15 such such ADJ bracis-19054 129 16 as as ADP bracis-19054 129 17 bow bow NOUN bracis-19054 129 18 or or CCONJ bracis-19054 129 19 document document NOUN bracis-19054 129 20 embedding embed VERB bracis-19054 129 21 [ [ X bracis-19054 129 22 1 1 NUM bracis-19054 129 23 ] ] PUNCT bracis-19054 129 24 must must AUX bracis-19054 129 25 be be AUX bracis-19054 129 26 adopted adopt VERB bracis-19054 129 27 to to PART bracis-19054 129 28 transform transform VERB bracis-19054 129 29 news news NOUN bracis-19054 129 30 into into ADP bracis-19054 129 31 structured structured ADJ bracis-19054 129 32 data datum NOUN bracis-19054 129 33 ( ( PUNCT bracis-19054 129 34 fig fig NOUN bracis-19054 129 35 . . PUNCT bracis-19054 129 36     SPACE bracis-19054 130 1 1 1 NUM bracis-19054 130 2 stage stage NOUN bracis-19054 130 3     SPACE bracis-19054 130 4 2 2 NUM bracis-19054 130 5 ) ) PUNCT bracis-19054 130 6 . . PUNCT bracis-19054 131 1 next next ADV bracis-19054 131 2 , , PUNCT bracis-19054 131 3 pu pu PROPN bracis-19054 131 4 - - PUNCT bracis-19054 131 5 lp lp PROPN bracis-19054 131 6 builds build VERB bracis-19054 131 7 an an DET bracis-19054 131 8 adjacency adjacency NOUN bracis-19054 131 9 matrix matrix NOUN bracis-19054 131 10 with with ADP bracis-19054 131 11 the the DET bracis-19054 131 12 complete complete ADJ bracis-19054 131 13 set set NOUN bracis-19054 131 14 of of ADP bracis-19054 131 15 examples example NOUN bracis-19054 131 16 , , PUNCT bracis-19054 131 17 \(\mathcal \(\mathcal ADJ bracis-19054 131 18 { { PUNCT bracis-19054 131 19 d}\ d}\ PROPN bracis-19054 131 20 ) ) PUNCT bracis-19054 131 21 , , PUNCT bracis-19054 131 22 detailed detail VERB bracis-19054 131 23 below below ADV bracis-19054 131 24 . . PUNCT bracis-19054 132 1 4.2 4.2 NUM bracis-19054 132 2 k k NOUN bracis-19054 132 3 - - PUNCT bracis-19054 132 4 nn nn X bracis-19054 132 5 matrix matrix NOUN bracis-19054 132 6 , , PUNCT bracis-19054 132 7 katz katz PROPN bracis-19054 132 8 index index PROPN bracis-19054 132 9 and and CCONJ bracis-19054 132 10     SPACE bracis-19054 132 11 sets set VERB bracis-19054 132 12 extraction extraction NOUN bracis-19054 132 13 the the DET bracis-19054 132 14 representation representation NOUN bracis-19054 132 15 model model NOUN bracis-19054 132 16 and and CCONJ bracis-19054 132 17 a a DET bracis-19054 132 18 chosen choose VERB bracis-19054 132 19 distance distance NOUN bracis-19054 132 20 metric metric NOUN bracis-19054 132 21 are be AUX bracis-19054 132 22 used use VERB bracis-19054 132 23 to to PART bracis-19054 132 24 calculate calculate VERB bracis-19054 132 25 an an DET bracis-19054 132 26 adjacency adjacency NOUN bracis-19054 132 27 matrix matrix NOUN bracis-19054 132 28 . . PUNCT bracis-19054 133 1 in in ADP bracis-19054 133 2 this this DET bracis-19054 133 3 matrix matrix NOUN bracis-19054 133 4 , , PUNCT bracis-19054 133 5 news news NOUN bracis-19054 133 6 with with ADP bracis-19054 133 7 similar similar ADJ bracis-19054 133 8 content content NOUN bracis-19054 133 9 has have VERB bracis-19054 133 10 a a DET bracis-19054 133 11 low low ADJ bracis-19054 133 12 distance distance NOUN bracis-19054 133 13 between between ADP bracis-19054 133 14 them they PRON bracis-19054 133 15 . . PUNCT bracis-19054 134 1 the the DET bracis-19054 134 2 adjacency adjacency NOUN bracis-19054 134 3 matrix matrix NOUN bracis-19054 134 4 is be AUX bracis-19054 134 5 used use VERB bracis-19054 134 6 as as ADP bracis-19054 134 7 a a DET bracis-19054 134 8 basis basis NOUN bracis-19054 134 9 for for ADP bracis-19054 134 10 building build VERB bracis-19054 134 11 a a DET bracis-19054 134 12 k k NOUN bracis-19054 134 13 - - PUNCT bracis-19054 134 14 nearest near ADJ bracis-19054 134 15 neighbors neighbor NOUN bracis-19054 134 16 ( ( PUNCT bracis-19054 134 17 k k X bracis-19054 134 18 - - PUNCT bracis-19054 134 19 nn nn ADJ bracis-19054 134 20 ) ) PUNCT bracis-19054 134 21 matrix matrix NOUN bracis-19054 134 22 , , PUNCT bracis-19054 134 23 called call VERB bracis-19054 134 24 a a PRON bracis-19054 134 25 , , PUNCT bracis-19054 134 26 so so SCONJ bracis-19054 134 27 that that SCONJ bracis-19054 134 28 \(a_{i \(a_{i ADP bracis-19054 134 29 , , PUNCT bracis-19054 134 30 j j NOUN bracis-19054 134 31 } } PUNCT bracis-19054 134 32 = = SYM bracis-19054 134 33 1\ 1\ NUM bracis-19054 134 34 ) ) PUNCT bracis-19054 134 35 if if SCONJ bracis-19054 134 36 the the DET bracis-19054 134 37 news news NOUN bracis-19054 134 38 \(d_j\ \(d_j\ NOUN bracis-19054 134 39 ) ) PUNCT bracis-19054 134 40 is be AUX bracis-19054 134 41 one one NUM bracis-19054 134 42 of of ADP bracis-19054 134 43 the the DET bracis-19054 134 44 k k PROPN bracis-19054 134 45 nearest near ADJ bracis-19054 134 46 neighbors neighbor NOUN bracis-19054 134 47 to to ADP bracis-19054 134 48 the the DET bracis-19054 134 49 news news NOUN bracis-19054 134 50 \(d_i\ \(d_i\ NOUN bracis-19054 134 51 ) ) PUNCT bracis-19054 134 52 , , PUNCT bracis-19054 134 53 and and CCONJ bracis-19054 134 54 \(a_{i \(a_{i PROPN bracis-19054 134 55 , , PUNCT bracis-19054 134 56 j j NOUN bracis-19054 134 57 } } PUNCT bracis-19054 134 58 = = SYM bracis-19054 134 59 0\ 0\ PROPN bracis-19054 134 60 ) ) PUNCT bracis-19054 134 61 otherwise otherwise ADV bracis-19054 134 62 ( ( PUNCT bracis-19054 134 63 fig fig NOUN bracis-19054 134 64 . . PUNCT bracis-19054 134 65     SPACE bracis-19054 135 1 1 1 NUM bracis-19054 135 2 stage stage NOUN bracis-19054 135 3 3 3 NUM bracis-19054 135 4 ) ) PUNCT bracis-19054 135 5 . . PUNCT bracis-19054 136 1 through through ADP bracis-19054 136 2 the the DET bracis-19054 136 3 k k PROPN bracis-19054 136 4 - - PUNCT bracis-19054 136 5 nn nn ADJ bracis-19054 136 6 matrix matrix NOUN bracis-19054 136 7 , , PUNCT bracis-19054 136 8 a a DET bracis-19054 136 9 heterogeneous heterogeneous ADJ bracis-19054 136 10 network network NOUN bracis-19054 136 11 \(\mathcal \(\mathcal ADJ bracis-19054 136 12 { { PUNCT bracis-19054 136 13 n}\ n}\ NOUN bracis-19054 136 14 ) ) PUNCT bracis-19054 136 15 is be AUX bracis-19054 136 16 also also ADV bracis-19054 136 17 created create VERB bracis-19054 136 18 . . PUNCT bracis-19054 137 1 the the DET bracis-19054 137 2 heterogeneous heterogeneous ADJ bracis-19054 137 3 network network NOUN bracis-19054 137 4 can can AUX bracis-19054 137 5 be be AUX bracis-19054 137 6 defined define VERB bracis-19054 137 7 as as ADP bracis-19054 137 8 a a DET bracis-19054 137 9 triple triple ADJ bracis-19054 137 10 \(\mathcal \(\mathcal ADJ bracis-19054 137 11 { { PUNCT bracis-19054 137 12 n n CCONJ bracis-19054 137 13 } } PUNCT bracis-19054 137 14 = = PUNCT bracis-19054 137 15 \langle \langle PROPN bracis-19054 137 16 \mathcal \mathcal INTJ bracis-19054 137 17 { { PUNCT bracis-19054 137 18 o},\mathcal o},\mathcal ADJ bracis-19054 137 19 { { PUNCT bracis-19054 137 20 r},\mathcal r},\mathcal ADJ bracis-19054 137 21 { { PUNCT bracis-19054 137 22 w w NOUN bracis-19054 137 23 } } PUNCT bracis-19054 137 24 \rangle \rangle PROPN bracis-19054 137 25 \ \ NOUN bracis-19054 137 26 ) ) PUNCT bracis-19054 137 27 , , PUNCT bracis-19054 137 28 in in ADP bracis-19054 137 29 which which PRON bracis-19054 137 30 \(\mathcal \(\mathcal ADJ bracis-19054 137 31 { { PUNCT bracis-19054 137 32 o}\ o}\ NOUN bracis-19054 137 33 ) ) PUNCT bracis-19054 137 34 is be AUX bracis-19054 137 35 the the DET bracis-19054 137 36 set set NOUN bracis-19054 137 37 of of ADP bracis-19054 137 38 objects object NOUN bracis-19054 137 39 , , PUNCT bracis-19054 137 40 \(\mathcal \(\mathcal ADJ bracis-19054 137 41 { { PUNCT bracis-19054 137 42 r}\ r}\ NUM bracis-19054 137 43 ) ) PUNCT bracis-19054 137 44 is be AUX bracis-19054 137 45 the the DET bracis-19054 137 46 set set NOUN bracis-19054 137 47 of of ADP bracis-19054 137 48 relations relation NOUN bracis-19054 137 49 between between ADP bracis-19054 137 50 objects object NOUN bracis-19054 137 51 , , PUNCT bracis-19054 137 52 and and CCONJ bracis-19054 137 53 \(\mathcal \(\mathcal ADJ bracis-19054 137 54 { { PUNCT bracis-19054 137 55 w}\ w}\ PROPN bracis-19054 137 56 ) ) PUNCT bracis-19054 137 57 is be AUX bracis-19054 137 58 the the DET bracis-19054 137 59 set set NOUN bracis-19054 137 60 of of ADP bracis-19054 137 61 weights weight NOUN bracis-19054 137 62 of of ADP bracis-19054 137 63 these these DET bracis-19054 137 64 relations relation NOUN bracis-19054 137 65 . . PUNCT bracis-19054 138 1 given give VERB bracis-19054 138 2 two two NUM bracis-19054 138 3 pairs pair NOUN bracis-19054 138 4 of of ADP bracis-19054 138 5 objects object NOUN bracis-19054 138 6 \(o_i \(o_i VERB bracis-19054 138 7 , , PUNCT bracis-19054 138 8 o_j o_j PUNCT bracis-19054 138 9 \in \in PROPN bracis-19054 138 10 \mathcal \mathcal PROPN bracis-19054 138 11 { { PUNCT bracis-19054 138 12 o}\ o}\ PROPN bracis-19054 138 13 ) ) PUNCT bracis-19054 138 14 , , PUNCT bracis-19054 138 15 the the DET bracis-19054 138 16 relationship relationship NOUN bracis-19054 138 17 between between ADP bracis-19054 138 18 them they PRON bracis-19054 138 19 is be AUX bracis-19054 138 20 \(r_{o_i \(r_{o_i NOUN bracis-19054 138 21 , , PUNCT bracis-19054 138 22 o_j}\ o_j}\ NOUN bracis-19054 138 23 ) ) PUNCT bracis-19054 138 24 and and CCONJ bracis-19054 138 25 the the DET bracis-19054 138 26 weight weight NOUN bracis-19054 138 27 of of ADP bracis-19054 138 28 a a DET bracis-19054 138 29 relation relation NOUN bracis-19054 138 30 \(r_{o_i \(r_{o_i NOUN bracis-19054 138 31 , , PUNCT bracis-19054 138 32 o_j}\ o_j}\ PROPN bracis-19054 138 33 ) ) PUNCT bracis-19054 138 34 is be AUX bracis-19054 138 35 given give VERB bracis-19054 138 36 by by ADP bracis-19054 138 37 \(w_{o_i \(w_{o_i NOUN bracis-19054 138 38 , , PUNCT bracis-19054 138 39 o_j o_j SYM bracis-19054 138 40 } } PUNCT bracis-19054 138 41 \forall \forall DET bracis-19054 138 42 o_i o_i NOUN bracis-19054 138 43 , , PUNCT bracis-19054 138 44 o_j o_j SYM bracis-19054 138 45 \in \in PROPN bracis-19054 138 46 \mathcal \mathcal PROPN bracis-19054 138 47 { { PUNCT bracis-19054 138 48 o}\ o}\ PROPN bracis-19054 138 49 ) ) PUNCT bracis-19054 138 50 . . PUNCT bracis-19054 139 1 in in ADP bracis-19054 139 2 \(\mathcal \(\mathcal ADJ bracis-19054 139 3 { { PUNCT bracis-19054 139 4 n}\ n}\ NUM bracis-19054 139 5 ) ) PUNCT bracis-19054 139 6 , , PUNCT bracis-19054 139 7 objects object NOUN bracis-19054 139 8 are be AUX bracis-19054 139 9 news news NOUN bracis-19054 139 10 , , PUNCT bracis-19054 139 11 the the DET bracis-19054 139 12 relationships relationship NOUN bracis-19054 139 13 between between ADP bracis-19054 139 14 news news NOUN bracis-19054 139 15 are be AUX bracis-19054 139 16 created create VERB bracis-19054 139 17 according accord VERB bracis-19054 139 18 to to ADP bracis-19054 139 19 the the DET bracis-19054 139 20 k k PROPN bracis-19054 139 21 - - PUNCT bracis-19054 139 22 nn nn X bracis-19054 139 23 matrix matrix NOUN bracis-19054 139 24 , , PUNCT bracis-19054 139 25 and and CCONJ bracis-19054 139 26 the the DET bracis-19054 139 27 weights weight NOUN bracis-19054 139 28 \(w_{o_i \(w_{o_i NOUN bracis-19054 139 29 , , PUNCT bracis-19054 139 30 o_j}\ o_j}\ PROPN bracis-19054 139 31 ) ) PUNCT bracis-19054 139 32 are be AUX bracis-19054 139 33 the the DET bracis-19054 139 34 cosine cosine NOUN bracis-19054 139 35 similarity similarity NOUN bracis-19054 139 36 between between ADP bracis-19054 139 37 news news NOUN bracis-19054 139 38 . . PUNCT bracis-19054 140 1 although although SCONJ bracis-19054 140 2 the the DET bracis-19054 140 3 cosine cosine NOUN bracis-19054 140 4 is be AUX bracis-19054 140 5 a a DET bracis-19054 140 6 good good ADJ bracis-19054 140 7 local local ADJ bracis-19054 140 8 similarity similarity NOUN bracis-19054 140 9 measure measure NOUN bracis-19054 140 10 , , PUNCT bracis-19054 140 11 its its PRON bracis-19054 140 12 performance performance NOUN bracis-19054 140 13 as as ADP bracis-19054 140 14 a a DET bracis-19054 140 15 global global ADJ bracis-19054 140 16 similarity similarity NOUN bracis-19054 140 17 measure measure NOUN bracis-19054 140 18 is be AUX bracis-19054 140 19 not not PART bracis-19054 140 20 good good ADJ bracis-19054 140 21 enough enough ADV bracis-19054 140 22 . . PUNCT bracis-19054 141 1 to to PART bracis-19054 141 2 consider consider VERB bracis-19054 141 3 local local ADJ bracis-19054 141 4 paths path NOUN bracis-19054 141 5 in in ADP bracis-19054 141 6 the the DET bracis-19054 141 7 news news NOUN bracis-19054 141 8 network network NOUN bracis-19054 141 9 that that PRON bracis-19054 141 10 involve involve VERB bracis-19054 141 11 neighboring neighboring NOUN bracis-19054 141 12 news news NOUN bracis-19054 141 13 in in ADP bracis-19054 141 14 common common ADJ bracis-19054 141 15 , , PUNCT bracis-19054 141 16 [ [ X bracis-19054 141 17 9 9 NUM bracis-19054 141 18 ] ] X bracis-19054 141 19 propose propose VERB bracis-19054 141 20 using use VERB bracis-19054 141 21 the the DET bracis-19054 141 22 katz katz PROPN bracis-19054 141 23 index index PROPN bracis-19054 141 24 . . PUNCT bracis-19054 142 1 the the DET bracis-19054 142 2 basic basic ADJ bracis-19054 142 3 idea idea NOUN bracis-19054 142 4 is be AUX bracis-19054 142 5 that that SCONJ bracis-19054 142 6 if if SCONJ bracis-19054 142 7 two two NUM bracis-19054 142 8 news news NOUN bracis-19054 142 9 have have VERB bracis-19054 142 10 many many ADJ bracis-19054 142 11 neighbors neighbor NOUN bracis-19054 142 12 in in ADP bracis-19054 142 13 common common ADJ bracis-19054 142 14 , , PUNCT bracis-19054 142 15 they they PRON bracis-19054 142 16 are be AUX bracis-19054 142 17 likely likely ADJ bracis-19054 142 18 to to PART bracis-19054 142 19 be be AUX bracis-19054 142 20 of of ADP bracis-19054 142 21 the the DET bracis-19054 142 22 same same ADJ bracis-19054 142 23 class class NOUN bracis-19054 142 24 . . PUNCT bracis-19054 143 1 katz katz PROPN bracis-19054 143 2 index index PROPN bracis-19054 143 3 is be AUX bracis-19054 143 4 a a DET bracis-19054 143 5 global global ADJ bracis-19054 143 6 similarity similarity NOUN bracis-19054 143 7 measure measure NOUN bracis-19054 143 8 that that PRON bracis-19054 143 9 calculates calculate VERB bracis-19054 143 10 the the DET bracis-19054 143 11 similarity similarity NOUN bracis-19054 143 12 between between ADP bracis-19054 143 13 pairs pair NOUN bracis-19054 143 14 of of ADP bracis-19054 143 15 nodes node NOUN bracis-19054 143 16 considering consider VERB bracis-19054 143 17 all all DET bracis-19054 143 18 possible possible ADJ bracis-19054 143 19 paths path NOUN bracis-19054 143 20 in in ADP bracis-19054 143 21 a a DET bracis-19054 143 22 graph graph NOUN bracis-19054 143 23 that that PRON bracis-19054 143 24 connect connect VERB bracis-19054 143 25 them they PRON bracis-19054 143 26 . . PUNCT bracis-19054 144 1 thus thus ADV bracis-19054 144 2 : : PUNCT bracis-19054 144 3 $ $ SYM bracis-19054 144 4 $ $ SYM bracis-19054 144 5 \begin{aligned \begin{aligne VERB bracis-19054 144 6 } } PUNCT bracis-19054 144 7 sim(d_i sim(d_i NOUN bracis-19054 144 8 , , PUNCT bracis-19054 144 9 d_j d_j NUM bracis-19054 144 10 ) ) PUNCT bracis-19054 144 11 = = SYM bracis-19054 144 12 \sum \sum NOUN bracis-19054 144 13 ^{\infty ^{\infty ADJ bracis-19054 144 14 } } PUNCT bracis-19054 144 15 _ _ PRON bracis-19054 144 16 { { PUNCT bracis-19054 144 17 h=1 h=1 NOUN bracis-19054 144 18 } } PUNCT bracis-19054 144 19 \alpha \alpha ADP bracis-19054 144 20 ^h ^h X bracis-19054 144 21 \cdot \cdot NOUN bracis-19054 144 22 |\textit{path}_{d_i |\textit{path}_{d_i NOUN bracis-19054 144 23 , , PUNCT bracis-19054 144 24 d_j}^{}| d_j}^{}| PROPN bracis-19054 145 1 = = PUNCT bracis-19054 145 2 \alpha \alpha PROPN bracis-19054 145 3 a_{i a_{i PROPN bracis-19054 145 4 , , PUNCT bracis-19054 145 5 j j NOUN bracis-19054 145 6 } } PUNCT bracis-19054 145 7 + + CCONJ bracis-19054 145 8 \alpha \alpha ADP bracis-19054 145 9 ^2 ^2 PUNCT bracis-19054 145 10 ( ( PUNCT bracis-19054 145 11 a^2)_{i a^2)_{i PROPN bracis-19054 145 12 , , PUNCT bracis-19054 145 13 j j NOUN bracis-19054 145 14 } } PUNCT bracis-19054 145 15 + + CCONJ bracis-19054 145 16 \alpha \alpha PROPN bracis-19054 145 17 ^3 ^3 PRON bracis-19054 145 18 ( ( PUNCT bracis-19054 145 19 a^3)_{i a^3)_{i PROPN bracis-19054 145 20 , , PUNCT bracis-19054 145 21 j j NOUN bracis-19054 145 22 } } PUNCT bracis-19054 145 23 + + CCONJ bracis-19054 145 24 ... ... PUNCT bracis-19054 145 25 , , PUNCT bracis-19054 145 26 \end{aligned}$$ \end{aligned}$$ X bracis-19054 145 27 ( ( PUNCT bracis-19054 145 28 1 1 NUM bracis-19054 145 29 ) ) PUNCT bracis-19054 145 30 in in ADP bracis-19054 145 31 which which PRON bracis-19054 145 32 \(\alpha \(\alpha X bracis-19054 145 33 \ \ NOUN bracis-19054 145 34 ) ) PUNCT bracis-19054 145 35 is be AUX bracis-19054 145 36 a a DET bracis-19054 145 37 parameter parameter NOUN bracis-19054 145 38 that that PRON bracis-19054 145 39 controls control VERB bracis-19054 145 40 the the DET bracis-19054 145 41 influence influence NOUN bracis-19054 145 42 of of ADP bracis-19054 145 43 paths path NOUN bracis-19054 145 44 . . PUNCT bracis-19054 146 1 with with ADP bracis-19054 146 2 a a DET bracis-19054 146 3 very very ADV bracis-19054 146 4 small small ADJ bracis-19054 146 5 \(\alpha \(\alpha X bracis-19054 146 6 \ \ NOUN bracis-19054 146 7 ) ) PUNCT bracis-19054 146 8 , , PUNCT bracis-19054 146 9 the the DET bracis-19054 146 10 long long ADJ bracis-19054 146 11 paths path NOUN bracis-19054 146 12 contribute contribute VERB bracis-19054 146 13 very very ADV bracis-19054 146 14 little little ADJ bracis-19054 146 15 . . PUNCT bracis-19054 147 1 in in ADP bracis-19054 147 2 the the DET bracis-19054 147 3 katz katz PROPN bracis-19054 147 4 index index PROPN bracis-19054 147 5 , , PUNCT bracis-19054 147 6 when when SCONJ bracis-19054 147 7 \(\alpha \(\alpha ADV bracis-19054 147 8 < < X bracis-19054 147 9 1/\epsilon 1/\epsilon NUM bracis-19054 147 10 \ \ NOUN bracis-19054 147 11 ) ) PUNCT bracis-19054 147 12 , , PUNCT bracis-19054 147 13 being be AUX bracis-19054 147 14 \(\epsilon \(\epsilon NOUN bracis-19054 147 15 \ \ NOUN bracis-19054 147 16 ) ) PUNCT bracis-19054 147 17 the the DET bracis-19054 147 18 biggest big ADJ bracis-19054 147 19 eigenvalue eigenvalue NOUN bracis-19054 147 20 for for ADP bracis-19054 147 21 the the DET bracis-19054 147 22 matrix matrix NOUN bracis-19054 147 23 a a DET bracis-19054 147 24 , , PUNCT bracis-19054 147 25 eq eq NOUN bracis-19054 147 26 . . PROPN bracis-19054 147 27     SPACE bracis-19054 147 28 1 1 NUM bracis-19054 147 29 converges converge NOUN bracis-19054 147 30 and and CCONJ bracis-19054 147 31 can can AUX bracis-19054 147 32 be be AUX bracis-19054 147 33 calculated calculate VERB bracis-19054 147 34 as as SCONJ bracis-19054 147 35 follows follow VERB bracis-19054 147 36 : : PUNCT bracis-19054 147 37 \(s \(s PROPN bracis-19054 147 38 = = SYM bracis-19054 147 39 ( ( PUNCT bracis-19054 147 40 i i PRON bracis-19054 147 41 \alpha \alpha ADP bracis-19054 147 42 a)^{-1 a)^{-1 NOUN bracis-19054 147 43 } } PUNCT bracis-19054 147 44 i\ i\ PROPN bracis-19054 147 45 ) ) PUNCT bracis-19054 147 46 , , PUNCT bracis-19054 147 47 which which PRON bracis-19054 147 48 i i PRON bracis-19054 147 49 denotes denote VERB bracis-19054 147 50 the the DET bracis-19054 147 51 identity identity NOUN bracis-19054 147 52 matrix matrix NOUN bracis-19054 147 53 and and CCONJ bracis-19054 147 54 s s AUX bracis-19054 147 55 has have VERB bracis-19054 147 56 dimnesions dimnesion NOUN bracis-19054 147 57 \(|\mathcal \(|\mathcal PUNCT bracis-19054 147 58 { { PUNCT bracis-19054 147 59 d}| d}| NOUN bracis-19054 147 60 \times \times ADP bracis-19054 147 61 |\mathcal |\mathcal ADJ bracis-19054 147 62 { { PUNCT bracis-19054 147 63 d}|\ d}|\ NOUN bracis-19054 147 64 ) ) PUNCT bracis-19054 147 65 . . PUNCT bracis-19054 148 1 thus thus ADV bracis-19054 148 2 , , PUNCT bracis-19054 148 3 \(s_{i \(s_{i PROPN bracis-19054 148 4 , , PUNCT bracis-19054 148 5 j j NOUN bracis-19054 148 6 } } PUNCT bracis-19054 148 7 \in \in PROPN bracis-19054 148 8 \mathbb \mathbb PROPN bracis-19054 148 9 { { PUNCT bracis-19054 148 10 r}\ r}\ NUM bracis-19054 148 11 ) ) PUNCT bracis-19054 148 12 denotes denote VERB bracis-19054 148 13 the the DET bracis-19054 148 14 similarity similarity NOUN bracis-19054 148 15 between between ADP bracis-19054 148 16 the the DET bracis-19054 148 17 nodes node NOUN bracis-19054 148 18 \(d_i\ \(d_i\ NOUN bracis-19054 148 19 ) ) PUNCT bracis-19054 148 20 and and CCONJ bracis-19054 148 21 \(d_j\ \(d_j\ PROPN bracis-19054 148 22 ) ) PUNCT bracis-19054 148 23 according accord VERB bracis-19054 148 24 to to ADP bracis-19054 148 25 katz katz PROPN bracis-19054 148 26 index index PROPN bracis-19054 148 27 ( ( PUNCT bracis-19054 148 28 fig fig NOUN bracis-19054 148 29 . . PUNCT bracis-19054 148 30     SPACE bracis-19054 149 1 1 1 NUM bracis-19054 149 2 stage stage NOUN bracis-19054 149 3     SPACE bracis-19054 149 4 4 4 NUM bracis-19054 149 5 ) ) PUNCT bracis-19054 149 6 . . PUNCT bracis-19054 150 1 the the DET bracis-19054 150 2 labeled label VERB bracis-19054 150 3 news news NOUN bracis-19054 150 4 in in ADP bracis-19054 150 5 \(\mathcal \(\mathcal ADJ bracis-19054 150 6 { { PUNCT bracis-19054 150 7 d}^{+}\ d}^{+}\ NOUN bracis-19054 150 8 ) ) PUNCT bracis-19054 150 9 and and CCONJ bracis-19054 150 10 the the DET bracis-19054 150 11 similarity similarity NOUN bracis-19054 150 12 matrix matrix NOUN bracis-19054 150 13 s s NOUN bracis-19054 150 14 are be AUX bracis-19054 150 15 used use VERB bracis-19054 150 16 to to PART bracis-19054 150 17 infer infer VERB bracis-19054 150 18 two two NUM bracis-19054 150 19 sets set NOUN bracis-19054 150 20 : : PUNCT bracis-19054 150 21 the the DET bracis-19054 150 22 set set NOUN bracis-19054 150 23 of of ADP bracis-19054 150 24 reliable reliable ADJ bracis-19054 150 25 interest interest NOUN bracis-19054 150 26 news news PROPN bracis-19054 150 27 ri ri PROPN bracis-19054 150 28 and and CCONJ bracis-19054 150 29 reliable reliable ADJ bracis-19054 150 30 non non ADJ bracis-19054 150 31 - - ADJ bracis-19054 150 32 interest interest ADJ bracis-19054 150 33 news news PROPN bracis-19054 150 34 rn rn PROPN bracis-19054 150 35 ( ( PUNCT bracis-19054 150 36 fig fig NOUN bracis-19054 150 37 . . PUNCT bracis-19054 150 38     SPACE bracis-19054 150 39 1 1 NUM bracis-19054 150 40 stage stage NOUN bracis-19054 150 41 5 5 NUM bracis-19054 150 42 ) ) PUNCT bracis-19054 150 43 . . PUNCT bracis-19054 151 1 the the DET bracis-19054 151 2 set set NOUN bracis-19054 151 3 ri ri PROPN bracis-19054 151 4 contains contain VERB bracis-19054 151 5 news news NOUN bracis-19054 151 6 from from ADP bracis-19054 151 7 set set VERB bracis-19054 151 8 \(\mathcal \(\mathcal ADJ bracis-19054 151 9 { { PUNCT bracis-19054 151 10 d}^{u}\ d}^{u}\ NOUN bracis-19054 151 11 ) ) PUNCT bracis-19054 151 12 that that PRON bracis-19054 151 13 are be AUX bracis-19054 151 14 most most ADV bracis-19054 151 15 similar similar ADJ bracis-19054 151 16 to to ADP bracis-19054 151 17 examples example NOUN bracis-19054 151 18 from from ADP bracis-19054 151 19 \(\mathcal \(\mathcal ADJ bracis-19054 151 20 { { PUNCT bracis-19054 151 21 d}^{+}\ d}^{+}\ NOUN bracis-19054 151 22 ) ) PUNCT bracis-19054 151 23 , , PUNCT bracis-19054 151 24 and and CCONJ bracis-19054 151 25 rn rn PROPN bracis-19054 151 26 will will AUX bracis-19054 151 27 contain contain VERB bracis-19054 151 28 news news NOUN bracis-19054 151 29 from from ADP bracis-19054 151 30 \(\mathcal \(\mathcal ADJ bracis-19054 151 31 { { PUNCT bracis-19054 151 32 d}^{u d}^{u PROPN bracis-19054 151 33 } } PUNCT bracis-19054 151 34 ri\ ri\ NOUN bracis-19054 151 35 ) ) PUNCT bracis-19054 151 36 that that PRON bracis-19054 151 37 are be AUX bracis-19054 151 38 most most ADV bracis-19054 151 39 dissimilar dissimilar ADJ bracis-19054 151 40 to to ADP bracis-19054 151 41 the the DET bracis-19054 151 42 set set ADJ bracis-19054 151 43 \(\mathcal \(\mathcal ADJ bracis-19054 151 44 { { PUNCT bracis-19054 151 45 d}^{+ d}^{+ NOUN bracis-19054 151 46 } } PUNCT bracis-19054 151 47 \cup \cup VERB bracis-19054 151 48 ri\ ri\ NOUN bracis-19054 151 49 ) ) PUNCT bracis-19054 151 50 . . PUNCT bracis-19054 152 1 for for ADP bracis-19054 152 2 the the DET bracis-19054 152 3 inference inference NOUN bracis-19054 152 4 of of ADP bracis-19054 152 5 the the DET bracis-19054 152 6 ri ri PROPN bracis-19054 152 7 set set NOUN bracis-19054 152 8 , , PUNCT bracis-19054 152 9 an an DET bracis-19054 152 10 iterative iterative NOUN bracis-19054 152 11 method method NOUN bracis-19054 152 12 is be AUX bracis-19054 152 13 applied apply VERB bracis-19054 152 14 . . PUNCT bracis-19054 153 1 the the DET bracis-19054 153 2 task task NOUN bracis-19054 153 3 is be AUX bracis-19054 153 4 divided divide VERB bracis-19054 153 5 into into ADP bracis-19054 153 6 m m NOUN bracis-19054 153 7 iterations iteration NOUN bracis-19054 153 8 . . PUNCT bracis-19054 154 1 in in ADP bracis-19054 154 2 each each PRON bracis-19054 154 3 of of ADP bracis-19054 154 4 the the DET bracis-19054 154 5 m m NOUN bracis-19054 154 6 iterations iteration NOUN bracis-19054 154 7 , , PUNCT bracis-19054 154 8 the the DET bracis-19054 154 9 total total ADJ bracis-19054 154 10 number number NOUN bracis-19054 154 11 of of ADP bracis-19054 154 12 reliable reliable ADJ bracis-19054 154 13 interest interest NOUN bracis-19054 154 14 news news NOUN bracis-19054 154 15 that that PRON bracis-19054 154 16 will will AUX bracis-19054 154 17 be be AUX bracis-19054 154 18 extracted extract VERB bracis-19054 154 19 is be AUX bracis-19054 154 20 \((\lambda \((\lambda NOUN bracis-19054 154 21 /m /m PUNCT bracis-19054 154 22 ) ) PUNCT bracis-19054 154 23 \times \times ADP bracis-19054 154 24 |\mathcal |\mathcal ADJ bracis-19054 154 25 { { PUNCT bracis-19054 154 26 d}^{+}|\ d}^{+}|\ NOUN bracis-19054 154 27 ) ) PUNCT bracis-19054 154 28 , , PUNCT bracis-19054 154 29 which which PRON bracis-19054 154 30 \(\lambda \(\lambda NOUN bracis-19054 154 31 \ \ NOUN bracis-19054 154 32 ) ) PUNCT bracis-19054 154 33 controls control VERB bracis-19054 154 34 the the DET bracis-19054 154 35 set set NOUN bracis-19054 154 36 ’s ’s PART bracis-19054 154 37 size size NOUN bracis-19054 154 38 . . PUNCT bracis-19054 155 1 the the DET bracis-19054 155 2 news news NOUN bracis-19054 155 3 in in ADP bracis-19054 155 4 \(\mathcal \(\mathcal ADJ bracis-19054 155 5 { { PUNCT bracis-19054 155 6 d}^{u}\ d}^{u}\ NOUN bracis-19054 155 7 ) ) PUNCT bracis-19054 155 8 are be AUX bracis-19054 155 9 ranked rank VERB bracis-19054 155 10 according accord VERB bracis-19054 155 11 to to ADP bracis-19054 155 12 their their PRON bracis-19054 155 13 average average ADJ bracis-19054 155 14 similarities similarity NOUN bracis-19054 155 15 for for ADP bracis-19054 155 16 all all DET bracis-19054 155 17 news news NOUN bracis-19054 155 18 in in ADP bracis-19054 155 19 \(\mathcal \(\mathcal ADJ bracis-19054 155 20 { { PUNCT bracis-19054 155 21 d}^{+}\ d}^{+}\ NOUN bracis-19054 155 22 ) ) PUNCT bracis-19054 155 23 based base VERB bracis-19054 155 24 on on ADP bracis-19054 155 25 s. s. PROPN bracis-19054 155 26 the the DET bracis-19054 155 27 \((\lambda \((\lambda PROPN bracis-19054 155 28 /m)|\mathcal /m)|\mathcal NOUN bracis-19054 155 29 { { PUNCT bracis-19054 155 30 d}^{+}|\ d}^{+}|\ NOUN bracis-19054 155 31 ) ) PUNCT bracis-19054 155 32 most most ADV bracis-19054 155 33 similar similar ADJ bracis-19054 155 34 news news NOUN bracis-19054 155 35 are be AUX bracis-19054 155 36 taken take VERB bracis-19054 155 37 from from ADP bracis-19054 155 38 \(d^{u}\ \(d^{u}\ NOUN bracis-19054 155 39 ) ) PUNCT bracis-19054 155 40 and and CCONJ bracis-19054 155 41 added add VERB bracis-19054 155 42 into into ADP bracis-19054 155 43 \(ri'\ \(ri'\ NUM bracis-19054 155 44 ) ) PUNCT bracis-19054 155 45 . . PUNCT bracis-19054 156 1 at at ADP bracis-19054 156 2 the the DET bracis-19054 156 3 end end NOUN bracis-19054 156 4 of of ADP bracis-19054 156 5 each each DET bracis-19054 156 6 iteration iteration NOUN bracis-19054 156 7 , , PUNCT bracis-19054 156 8 ri ri PROPN bracis-19054 156 9 is be AUX bracis-19054 156 10 incremented incremente VERB bracis-19054 156 11 with with ADP bracis-19054 156 12 the the DET bracis-19054 156 13 elements element NOUN bracis-19054 156 14 in in ADP bracis-19054 156 15 \(ri'\ \(ri'\ NUM bracis-19054 156 16 ) ) PUNCT bracis-19054 156 17 ( ( PUNCT bracis-19054 156 18 \(ri \(ri NOUN bracis-19054 156 19 \leftarrow \leftarrow VERB bracis-19054 156 20 ri ri PROPN bracis-19054 156 21 \cup \cup PROPN bracis-19054 156 22 ri'\ ri'\ PROPN bracis-19054 156 23 ) ) PUNCT bracis-19054 156 24 ) ) PUNCT bracis-19054 157 1 [ [ X bracis-19054 157 2 9 9 NUM bracis-19054 157 3 ] ] PUNCT bracis-19054 157 4 . . PUNCT bracis-19054 158 1 for for ADP bracis-19054 158 2 the the DET bracis-19054 158 3 inference inference NOUN bracis-19054 158 4 of of ADP bracis-19054 158 5 the the DET bracis-19054 158 6 reliable reliable ADJ bracis-19054 158 7 non non ADJ bracis-19054 158 8 - - ADJ bracis-19054 158 9 interest interest ADJ bracis-19054 158 10 set set NOUN bracis-19054 158 11 , , PUNCT bracis-19054 158 12 news news NOUN bracis-19054 158 13 in in ADP bracis-19054 158 14 \(\mathcal \(\mathcal ADJ bracis-19054 158 15 { { PUNCT bracis-19054 158 16 d}^{u d}^{u PROPN bracis-19054 158 17 } } PUNCT bracis-19054 158 18 ri\ ri\ NOUN bracis-19054 158 19 ) ) PUNCT bracis-19054 158 20 are be AUX bracis-19054 158 21 ranked rank VERB bracis-19054 158 22 according accord VERB bracis-19054 158 23 to to ADP bracis-19054 158 24 their their PRON bracis-19054 158 25 average average ADJ bracis-19054 158 26 similarities similarity NOUN bracis-19054 158 27 ( ( PUNCT bracis-19054 158 28 based base VERB bracis-19054 158 29 on on ADP bracis-19054 158 30 s s NOUN bracis-19054 158 31 ) ) PUNCT bracis-19054 158 32 for for ADP bracis-19054 158 33 all all DET bracis-19054 158 34 news news NOUN bracis-19054 158 35 in in ADP bracis-19054 158 36 \(\mathcal \(\mathcal ADJ bracis-19054 158 37 { { PUNCT bracis-19054 158 38 d}^{+ d}^{+ NOUN bracis-19054 158 39 } } PUNCT bracis-19054 158 40 \cup \cup VERB bracis-19054 158 41 ri\ ri\ NOUN bracis-19054 158 42 ) ) PUNCT bracis-19054 158 43 . . PUNCT bracis-19054 159 1 the the DET bracis-19054 159 2 algorithm algorithm NOUN bracis-19054 159 3 extracts extract VERB bracis-19054 159 4 the the DET bracis-19054 159 5 \(|\mathcal \(|\mathcal ADJ bracis-19054 159 6 { { PUNCT bracis-19054 159 7 d}^{+ d}^{+ NOUN bracis-19054 159 8 } } PUNCT bracis-19054 159 9 \cup \cup VERB bracis-19054 159 10 ri|\ ri|\ NOUN bracis-19054 159 11 ) ) PUNCT bracis-19054 159 12 most most ADV bracis-19054 159 13 dissimilar dissimilar ADJ bracis-19054 159 14 news news NOUN bracis-19054 159 15 , , PUNCT bracis-19054 159 16 forming form VERB bracis-19054 159 17 the the DET bracis-19054 159 18 set set NOUN bracis-19054 159 19 rn rn PROPN bracis-19054 159 20 . . PROPN bracis-19054 160 1 after after ADP bracis-19054 160 2 getting get VERB bracis-19054 160 3 the the DET bracis-19054 160 4 set set NOUN bracis-19054 160 5 rn rn PROPN bracis-19054 160 6 , , PUNCT bracis-19054 160 7 the the DET bracis-19054 160 8 sets set NOUN bracis-19054 160 9 \(\mathcal \(\mathcal ADJ bracis-19054 160 10 { { PUNCT bracis-19054 160 11 d}^{+ d}^{+ NOUN bracis-19054 160 12 } } PUNCT bracis-19054 160 13 \cup \cup VERB bracis-19054 160 14 ri\ ri\ NOUN bracis-19054 160 15 ) ) PUNCT bracis-19054 160 16 , , PUNCT bracis-19054 160 17 rn rn PROPN bracis-19054 160 18 , , PUNCT bracis-19054 160 19 and and CCONJ bracis-19054 160 20 \(\mathcal \(\mathcal ADJ bracis-19054 160 21 { { PUNCT bracis-19054 160 22 d}^{u d}^{u PROPN bracis-19054 160 23 } } PUNCT bracis-19054 160 24 \leftarrow \leftarrow PROPN bracis-19054 160 25 ( ( PUNCT bracis-19054 160 26 \mathcal \mathcal PROPN bracis-19054 160 27 { { PUNCT bracis-19054 160 28 d}^{u d}^{u PROPN bracis-19054 160 29 } } PUNCT bracis-19054 160 30 ri ri NOUN bracis-19054 160 31 rn\ rn\ NOUN bracis-19054 160 32 ) ) PUNCT bracis-19054 160 33 ) ) PUNCT bracis-19054 160 34 are be AUX bracis-19054 160 35 used use VERB bracis-19054 160 36 as as ADP bracis-19054 160 37 input input NOUN bracis-19054 160 38 by by ADP bracis-19054 160 39 label label NOUN bracis-19054 160 40 propagation propagation NOUN bracis-19054 160 41 algorithms algorithm NOUN bracis-19054 160 42 based base VERB bracis-19054 160 43 on on ADP bracis-19054 160 44 transductive transductive ADJ bracis-19054 160 45 semi semi ADJ bracis-19054 160 46 - - ADJ bracis-19054 160 47 supervised supervised ADJ bracis-19054 160 48 learning learning NOUN bracis-19054 160 49 . . PUNCT bracis-19054 161 1 more more ADJ bracis-19054 161 2 details detail NOUN bracis-19054 161 3 about about ADP bracis-19054 161 4 the the DET bracis-19054 161 5 algorithm algorithm NOUN bracis-19054 161 6 can can AUX bracis-19054 161 7 be be AUX bracis-19054 161 8 seen see VERB bracis-19054 161 9 in in ADP bracis-19054 161 10 [ [ X bracis-19054 161 11 9 9 NUM bracis-19054 161 12 ] ] PUNCT bracis-19054 161 13 . . PUNCT bracis-19054 162 1 4.3 4.3 NUM bracis-19054 162 2 adding add VERB bracis-19054 162 3 features feature NOUN bracis-19054 162 4 in in ADP bracis-19054 162 5     SPACE bracis-19054 162 6 the the DET bracis-19054 162 7     SPACE bracis-19054 162 8 news news NOUN bracis-19054 162 9 network network NOUN bracis-19054 162 10 considering consider VERB bracis-19054 162 11 that that SCONJ bracis-19054 162 12 the the DET bracis-19054 162 13 pu pu PROPN bracis-19054 162 14 - - PUNCT bracis-19054 162 15 lp lp PROPN bracis-19054 162 16 algorithm algorithm NOUN bracis-19054 162 17 is be AUX bracis-19054 162 18 entirely entirely ADV bracis-19054 162 19 based base VERB bracis-19054 162 20 on on ADP bracis-19054 162 21 networks network NOUN bracis-19054 162 22 , , PUNCT bracis-19054 162 23 we we PRON bracis-19054 162 24 propose propose VERB bracis-19054 162 25 adding add VERB bracis-19054 162 26 new new ADJ bracis-19054 162 27 features feature NOUN bracis-19054 162 28 as as ADP bracis-19054 162 29 nodes node NOUN bracis-19054 162 30 into into ADP bracis-19054 162 31 the the DET bracis-19054 162 32 news news NOUN bracis-19054 162 33 network network NOUN bracis-19054 162 34 \(\mathcal \(\mathcal ADJ bracis-19054 162 35 { { PUNCT bracis-19054 162 36 n}\ n}\ NOUN bracis-19054 162 37 ) ) PUNCT bracis-19054 162 38 that that PRON bracis-19054 162 39 has have AUX bracis-19054 162 40 been be AUX bracis-19054 162 41 shown show VERB bracis-19054 162 42 to to PART bracis-19054 162 43 be be AUX bracis-19054 162 44 relevant relevant ADJ bracis-19054 162 45 in in ADP bracis-19054 162 46 [ [ X bracis-19054 162 47 2 2 NUM bracis-19054 162 48 , , PUNCT bracis-19054 162 49 17 17 NUM bracis-19054 162 50 ] ] PUNCT bracis-19054 162 51 in in ADP bracis-19054 162 52 news news NOUN bracis-19054 162 53 classification classification NOUN bracis-19054 162 54 : : PUNCT bracis-19054 162 55 incertainty incertainty ADJ bracis-19054 163 1 = = PUNCT bracis-19054 163 2 total total ADJ bracis-19054 163 3 number number NOUN bracis-19054 163 4 of of ADP bracis-19054 163 5 modal modal ADJ bracis-19054 163 6 verbs verb NOUN bracis-19054 163 7 and and CCONJ bracis-19054 163 8 passive passive ADJ bracis-19054 163 9 voice voice NOUN bracis-19054 163 10 non non ADJ bracis-19054 163 11 immediacy immediacy NOUN bracis-19054 163 12 = = PUNCT bracis-19054 163 13 total total NOUN bracis-19054 163 14 of of ADP bracis-19054 163 15 singular singular NOUN bracis-19054 163 16 first first ADJ bracis-19054 163 17 and and CCONJ bracis-19054 163 18 second second ADJ bracis-19054 163 19 personal personal ADJ bracis-19054 163 20 pronoun pronoun NOUN bracis-19054 163 21 emotiveness emotiveness NOUN bracis-19054 163 22 ( ( PUNCT bracis-19054 163 23 emo emo NOUN bracis-19054 163 24 ) ) PUNCT bracis-19054 163 25 = = PUNCT bracis-19054 163 26 \(\frac{\text \(\frac{\text PROPN bracis-19054 163 27 { { PUNCT bracis-19054 163 28 total total ADJ bracis-19054 163 29 number number NOUN bracis-19054 163 30 of of ADP bracis-19054 163 31 adjectives adjective NOUN bracis-19054 163 32 + + CCONJ bracis-19054 163 33 total total ADJ bracis-19054 163 34 number number NOUN bracis-19054 163 35 of of ADP bracis-19054 163 36 adverbs}}{\text adverbs}}{\text ADJ bracis-19054 163 37 { { PUNCT bracis-19054 163 38 total total ADJ bracis-19054 163 39 number number NOUN bracis-19054 163 40 of of ADP bracis-19054 163 41 nouns noun NOUN bracis-19054 163 42 + + CCONJ bracis-19054 163 43 total total ADJ bracis-19054 163 44 number number NOUN bracis-19054 163 45 of of ADP bracis-19054 163 46 verbs}}\ verbs}}\ NOUN bracis-19054 163 47 ) ) PUNCT bracis-19054 163 48 pausality pausality NOUN bracis-19054 163 49 ( ( PUNCT bracis-19054 163 50 pau pau NOUN bracis-19054 163 51 ) ) PUNCT bracis-19054 163 52 = = X bracis-19054 163 53 \(\frac{\text \(\frac{\text PROPN bracis-19054 163 54 { { PUNCT bracis-19054 163 55 total total ADJ bracis-19054 163 56 number number NOUN bracis-19054 163 57 of of ADP bracis-19054 163 58 punctuation punctuation NOUN bracis-19054 163 59 marks}}{\text marks}}{\text NOUN bracis-19054 163 60 { { PUNCT bracis-19054 163 61 total total ADJ bracis-19054 163 62 number number NOUN bracis-19054 163 63 of of ADP bracis-19054 163 64 sentences}}\ sentences}}\ PROPN bracis-19054 163 65 ) ) PUNCT bracis-19054 163 66 average average ADJ bracis-19054 163 67 words word NOUN bracis-19054 163 68 sentence sentence NOUN bracis-19054 163 69 ( ( PUNCT bracis-19054 163 70 avgws avgws PROPN bracis-19054 163 71 ) ) PUNCT bracis-19054 163 72 = = PUNCT bracis-19054 163 73 \(\frac{\text \(\frac{\text PROPN bracis-19054 163 74 { { PUNCT bracis-19054 163 75 total total ADJ bracis-19054 163 76 number number NOUN bracis-19054 163 77 of of ADP bracis-19054 163 78 words}}{\text words}}{\text PROPN bracis-19054 163 79 { { PUNCT bracis-19054 163 80 total total ADJ bracis-19054 163 81 number number NOUN bracis-19054 163 82 of of ADP bracis-19054 163 83 sentences}}\ sentences}}\ PROPN bracis-19054 163 84 ) ) PUNCT bracis-19054 163 85 among among ADP bracis-19054 163 86 the the DET bracis-19054 163 87 features feature NOUN bracis-19054 163 88 considered consider VERB bracis-19054 163 89 , , PUNCT bracis-19054 163 90 we we PRON bracis-19054 163 91 compute compute VERB bracis-19054 163 92 for for ADP bracis-19054 163 93 each each DET bracis-19054 163 94 dataset dataset NOUN bracis-19054 163 95 which which DET bracis-19054 163 96 ones one NOUN bracis-19054 163 97 had have VERB bracis-19054 163 98 the the DET bracis-19054 163 99 highest high ADJ bracis-19054 163 100 correlation correlation NOUN bracis-19054 163 101 with with ADP bracis-19054 163 102 the the DET bracis-19054 163 103 target target NOUN bracis-19054 163 104 attribute attribute NOUN bracis-19054 163 105 and and CCONJ bracis-19054 163 106 appeared appear VERB bracis-19054 163 107 in in ADP bracis-19054 163 108 more more ADJ bracis-19054 163 109 than than ADP bracis-19054 163 110 one one NUM bracis-19054 163 111 dataset dataset NOUN bracis-19054 163 112 . . PUNCT bracis-19054 164 1 the the DET bracis-19054 164 2 characteristics characteristic NOUN bracis-19054 164 3 chosen choose VERB bracis-19054 164 4 to to PART bracis-19054 164 5 be be AUX bracis-19054 164 6 included include VERB bracis-19054 164 7 in in ADP bracis-19054 164 8 the the DET bracis-19054 164 9 network network NOUN bracis-19054 164 10 were be AUX bracis-19054 164 11 : : PUNCT bracis-19054 164 12 pausality pausality NOUN bracis-19054 164 13 , , PUNCT bracis-19054 164 14 emotiveness emotiveness NOUN bracis-19054 164 15 , , PUNCT bracis-19054 164 16 and and CCONJ bracis-19054 164 17 average average ADJ bracis-19054 164 18 words word NOUN bracis-19054 164 19 per per ADP bracis-19054 164 20 sentence sentence NOUN bracis-19054 164 21 . . PUNCT bracis-19054 165 1 table table NOUN bracis-19054 165 2     SPACE bracis-19054 165 3 1 1 NUM bracis-19054 165 4 shows show VERB bracis-19054 165 5 the the DET bracis-19054 165 6 correlations correlation NOUN bracis-19054 165 7 values value NOUN bracis-19054 165 8 . . PUNCT bracis-19054 166 1 the the DET bracis-19054 166 2 features feature NOUN bracis-19054 166 3 are be AUX bracis-19054 166 4 added add VERB bracis-19054 166 5 to to ADP bracis-19054 166 6 the the DET bracis-19054 166 7 news news NOUN bracis-19054 166 8 network network NOUN bracis-19054 166 9 \(\mathcal \(\mathcal ADJ bracis-19054 166 10 { { PUNCT bracis-19054 166 11 n}\ n}\ NOUN bracis-19054 166 12 ) ) PUNCT bracis-19054 166 13 as as ADP bracis-19054 166 14 new new ADJ bracis-19054 166 15 unlabeled unlabele VERB bracis-19054 166 16 nodes node NOUN bracis-19054 166 17 \(f_j\ \(f_j\ PROPN bracis-19054 166 18 ) ) PUNCT bracis-19054 166 19 , , PUNCT bracis-19054 166 20 \(0 \(0 PROPN bracis-19054 166 21 < < X bracis-19054 166 22 j<\text j<\text NOUN bracis-19054 166 23 { { PUNCT bracis-19054 166 24 total total PROPN bracis-19054 166 25 features}\ features}\ PROPN bracis-19054 166 26 ) ) PUNCT bracis-19054 166 27 . . PUNCT bracis-19054 167 1 new new ADJ bracis-19054 167 2 edges edge NOUN bracis-19054 167 3 between between ADP bracis-19054 167 4 news news NOUN bracis-19054 167 5 and and CCONJ bracis-19054 167 6 the the DET bracis-19054 167 7 features feature NOUN bracis-19054 167 8 are be AUX bracis-19054 167 9 added add VERB bracis-19054 167 10 , , PUNCT bracis-19054 167 11 whose whose DET bracis-19054 167 12 weights weight NOUN bracis-19054 167 13 \(w_{d_i \(w_{d_i PROPN bracis-19054 167 14 , , PUNCT bracis-19054 167 15 f_j}\ f_j}\ NOUN bracis-19054 167 16 ) ) PUNCT bracis-19054 167 17 correspond correspond VERB bracis-19054 167 18 to to PART bracis-19054 167 19 normalized normalize VERB bracis-19054 167 20 values value NOUN bracis-19054 167 21 of of ADP bracis-19054 167 22 the the DET bracis-19054 167 23 feature feature NOUN bracis-19054 167 24 \(f_j\ \(f_j\ NOUN bracis-19054 167 25 ) ) PUNCT bracis-19054 167 26 for for ADP bracis-19054 167 27 the the DET bracis-19054 167 28 news news NOUN bracis-19054 167 29 \(d_i\ \(d_i\ NOUN bracis-19054 167 30 ) ) PUNCT bracis-19054 167 31 , , PUNCT bracis-19054 167 32 according accord VERB bracis-19054 167 33 to to ADP bracis-19054 167 34 the the DET bracis-19054 167 35 equation equation NOUN bracis-19054 167 36 : : PUNCT bracis-19054 167 37 . . PUNCT bracis-19054 168 1 we we PRON bracis-19054 168 2 also also ADV bracis-19054 168 3 considered consider VERB bracis-19054 168 4 other other ADJ bracis-19054 168 5 linguistic linguistic ADJ bracis-19054 168 6 features feature NOUN bracis-19054 168 7 extracted extract VERB bracis-19054 168 8 by by ADP bracis-19054 168 9 liwc liwc PROPN bracis-19054 169 1 [ [ X bracis-19054 169 2 10 10 NUM bracis-19054 169 3 ] ] X bracis-19054 169 4 , , PUNCT bracis-19054 169 5 such such ADJ bracis-19054 169 6 as as ADP bracis-19054 169 7 counting count VERB bracis-19054 169 8 personal personal ADJ bracis-19054 169 9 pronouns pronoun NOUN bracis-19054 169 10 , , PUNCT bracis-19054 169 11 prepositions preposition NOUN bracis-19054 169 12 , , PUNCT bracis-19054 169 13 verbs verb NOUN bracis-19054 169 14 , , PUNCT bracis-19054 169 15 nouns noun NOUN bracis-19054 169 16 , , PUNCT bracis-19054 169 17 adjectives adjective NOUN bracis-19054 169 18 and and CCONJ bracis-19054 169 19 quantitative quantitative ADJ bracis-19054 169 20 words word NOUN bracis-19054 169 21 , , PUNCT bracis-19054 169 22 use use NOUN bracis-19054 169 23 of of ADP bracis-19054 169 24 affective affective ADJ bracis-19054 169 25 words word NOUN bracis-19054 169 26 , , PUNCT bracis-19054 169 27 positive positive ADJ bracis-19054 169 28 and and CCONJ bracis-19054 169 29 negative negative ADJ bracis-19054 169 30 emotions emotion NOUN bracis-19054 169 31 , , PUNCT bracis-19054 169 32 and and CCONJ bracis-19054 169 33 anxiety anxiety NOUN bracis-19054 169 34 , , PUNCT bracis-19054 169 35 but but CCONJ bracis-19054 169 36 they they PRON bracis-19054 169 37 did do AUX bracis-19054 169 38 not not PART bracis-19054 169 39 demonstrated demonstrate VERB bracis-19054 169 40 to to PART bracis-19054 169 41 be be AUX bracis-19054 169 42 relevant relevant ADJ bracis-19054 169 43 in in ADP bracis-19054 169 44 our our PRON bracis-19054 169 45 proposal proposal NOUN bracis-19054 169 46 . . PUNCT bracis-19054 170 1 table table NOUN bracis-19054 170 2 1 1 NUM bracis-19054 170 3 . . PUNCT bracis-19054 171 1 correlations correlation NOUN bracis-19054 171 2 of of ADP bracis-19054 171 3 the the DET bracis-19054 171 4 main main ADJ bracis-19054 171 5 linguistic linguistic ADJ bracis-19054 171 6 features feature NOUN bracis-19054 171 7 considering consider VERB bracis-19054 171 8 all all DET bracis-19054 171 9 news news NOUN bracis-19054 171 10 datasets.full datasets.full NOUN bracis-19054 171 11 size size NOUN bracis-19054 171 12 table table NOUN bracis-19054 171 13 in in ADP bracis-19054 171 14 addition addition NOUN bracis-19054 171 15 to to ADP bracis-19054 171 16 the the DET bracis-19054 171 17 linguistic linguistic ADJ bracis-19054 171 18 characteristics characteristic NOUN bracis-19054 171 19 , , PUNCT bracis-19054 171 20 we we PRON bracis-19054 171 21 also also ADV bracis-19054 171 22 considered consider VERB bracis-19054 171 23 terms term NOUN bracis-19054 171 24 as as ADP bracis-19054 171 25 network network NOUN bracis-19054 171 26 nodes node NOUN bracis-19054 171 27 since since SCONJ bracis-19054 171 28 news news NOUN bracis-19054 171 29 terms term NOUN bracis-19054 171 30 are be AUX bracis-19054 171 31 widely widely ADV bracis-19054 171 32 used use VERB bracis-19054 171 33 in in ADP bracis-19054 171 34 researches research NOUN bracis-19054 171 35 for for ADP bracis-19054 171 36 discrimination discrimination NOUN bracis-19054 171 37 of of ADP bracis-19054 171 38 true true ADJ bracis-19054 171 39 and and CCONJ bracis-19054 171 40 false false ADJ bracis-19054 171 41 content content NOUN bracis-19054 171 42 . . PUNCT bracis-19054 172 1 they they PRON bracis-19054 172 2 also also ADV bracis-19054 172 3 demonstrated demonstrate VERB bracis-19054 172 4 to to PART bracis-19054 172 5 be be AUX bracis-19054 172 6 useful useful ADJ bracis-19054 172 7 in in ADP bracis-19054 172 8 label label NOUN bracis-19054 172 9 propagation propagation NOUN bracis-19054 172 10 for for ADP bracis-19054 172 11 text text NOUN bracis-19054 172 12 classification classification NOUN bracis-19054 172 13 [ [ X bracis-19054 172 14 12 12 NUM bracis-19054 172 15 ] ] PUNCT bracis-19054 172 16 . . PUNCT bracis-19054 173 1 to to PART bracis-19054 173 2 select select VERB bracis-19054 173 3 terms term NOUN bracis-19054 173 4 , , PUNCT bracis-19054 173 5 stopwords stopword NOUN bracis-19054 173 6 were be AUX bracis-19054 173 7 removed remove VERB bracis-19054 173 8 and and CCONJ bracis-19054 173 9 the the DET bracis-19054 173 10 terms term NOUN bracis-19054 173 11 were be AUX bracis-19054 173 12 stemmed stem VERB bracis-19054 173 13 . . PUNCT bracis-19054 174 1 then then ADV bracis-19054 174 2 , , PUNCT bracis-19054 174 3 the the DET bracis-19054 174 4 single single ADJ bracis-19054 174 5 terms term NOUN bracis-19054 174 6 and and CCONJ bracis-19054 174 7 sequences sequence NOUN bracis-19054 174 8 of of ADP bracis-19054 174 9 two two NUM bracis-19054 174 10 terms term NOUN bracis-19054 174 11 , , PUNCT bracis-19054 174 12 i.e. i.e. X bracis-19054 174 13 , , PUNCT bracis-19054 174 14 unigrams unigram NOUN bracis-19054 174 15 and and CCONJ bracis-19054 174 16 bigrams bigram NOUN bracis-19054 174 17 , , PUNCT bracis-19054 174 18 are be AUX bracis-19054 174 19 selected select VERB bracis-19054 174 20 to to PART bracis-19054 174 21 be be AUX bracis-19054 174 22 network network NOUN bracis-19054 174 23 nodes node NOUN bracis-19054 174 24 if if SCONJ bracis-19054 174 25 they they PRON bracis-19054 174 26 respect respect VERB bracis-19054 174 27 a a DET bracis-19054 174 28 minimum minimum ADJ bracis-19054 174 29 document document NOUN bracis-19054 174 30 frequency frequency NOUN bracis-19054 174 31 of of ADP bracis-19054 174 32 2 2 NUM bracis-19054 174 33 documents document NOUN bracis-19054 174 34 and and CCONJ bracis-19054 174 35 their their PRON bracis-19054 174 36 term term NOUN bracis-19054 174 37 frequency frequency NOUN bracis-19054 174 38 inverse inverse NOUN bracis-19054 174 39 document document NOUN bracis-19054 174 40 frequency frequency NOUN bracis-19054 174 41 ( ( PUNCT bracis-19054 174 42 tf tf PROPN bracis-19054 174 43 - - PUNCT bracis-19054 174 44 idf idf NOUN bracis-19054 174 45 ) ) PUNCT bracis-19054 174 46 value value NOUN bracis-19054 174 47 in in ADP bracis-19054 174 48 a a DET bracis-19054 174 49 document document NOUN bracis-19054 174 50 is be AUX bracis-19054 174 51 above above ADP bracis-19054 174 52 a a DET bracis-19054 174 53 threshold threshold NOUN bracis-19054 174 54 \(\ell \(\ell NOUN bracis-19054 174 55 \ \ NOUN bracis-19054 174 56 ) ) PUNCT bracis-19054 174 57 . . PUNCT bracis-19054 175 1 table table NOUN bracis-19054 175 2 2 2 NUM bracis-19054 175 3 . . X bracis-19054 175 4 list list NOUN bracis-19054 175 5 of of ADP bracis-19054 175 6 characteristics characteristic NOUN bracis-19054 175 7 included include VERB bracis-19054 175 8 in in ADP bracis-19054 175 9 each each DET bracis-19054 175 10 proposed propose VERB bracis-19054 175 11 heterogeneous heterogeneous ADJ bracis-19054 175 12 networks.full networks.full PRON bracis-19054 175 13 size size NOUN bracis-19054 175 14 table table NOUN bracis-19054 175 15 twelve twelve NUM bracis-19054 175 16 different different ADJ bracis-19054 175 17 heterogeneous heterogeneous ADJ bracis-19054 175 18 networks network NOUN bracis-19054 175 19 were be AUX bracis-19054 175 20 created create VERB bracis-19054 175 21 , , PUNCT bracis-19054 175 22 considering consider VERB bracis-19054 175 23 the the DET bracis-19054 175 24 combination combination NOUN bracis-19054 175 25 of of ADP bracis-19054 175 26 all all DET bracis-19054 175 27 features feature NOUN bracis-19054 175 28 . . PUNCT bracis-19054 176 1 table table NOUN bracis-19054 176 2     SPACE bracis-19054 176 3 2 2 NUM bracis-19054 176 4 presents present VERB bracis-19054 176 5 the the DET bracis-19054 176 6 networks network NOUN bracis-19054 176 7 and and CCONJ bracis-19054 176 8 features feature NOUN bracis-19054 176 9 included include VERB bracis-19054 176 10 in in ADP bracis-19054 176 11 each each DET bracis-19054 176 12 one one NUM bracis-19054 176 13 . . PUNCT bracis-19054 177 1 after after ADP bracis-19054 177 2 adding add VERB bracis-19054 177 3 all all DET bracis-19054 177 4 the the DET bracis-19054 177 5 features feature NOUN bracis-19054 177 6 as as ADP bracis-19054 177 7 nodes node NOUN bracis-19054 177 8 into into ADP bracis-19054 177 9 the the DET bracis-19054 177 10 network network NOUN bracis-19054 177 11 \(\mathcal \(\mathcal PUNCT bracis-19054 177 12 { { PUNCT bracis-19054 177 13 n}\ n}\ NUM bracis-19054 177 14 ) ) PUNCT bracis-19054 177 15 , , PUNCT bracis-19054 177 16 we we PRON bracis-19054 177 17 perform perform VERB bracis-19054 177 18 the the DET bracis-19054 177 19 normalization normalization NOUN bracis-19054 177 20 of of ADP bracis-19054 177 21 network network NOUN bracis-19054 177 22 relations relation NOUN bracis-19054 177 23 in in ADP bracis-19054 177 24 order order NOUN bracis-19054 177 25 to to PART bracis-19054 177 26 mitigate mitigate VERB bracis-19054 177 27 possible possible ADJ bracis-19054 177 28 distortions distortion NOUN bracis-19054 177 29 due due ADJ bracis-19054 177 30 to to ADP bracis-19054 177 31 different different ADJ bracis-19054 177 32 ranges range NOUN bracis-19054 177 33 of of ADP bracis-19054 177 34 the the DET bracis-19054 177 35 different different ADJ bracis-19054 177 36 types type NOUN bracis-19054 177 37 of of ADP bracis-19054 177 38 relations relation NOUN bracis-19054 177 39 . . PUNCT bracis-19054 178 1 the the DET bracis-19054 178 2 relations relation NOUN bracis-19054 178 3 are be AUX bracis-19054 178 4 normalized normalize VERB bracis-19054 178 5 considering consider VERB bracis-19054 178 6 each each DET bracis-19054 178 7 relation relation NOUN bracis-19054 178 8 type type NOUN bracis-19054 178 9 . . PUNCT bracis-19054 179 1 thus thus ADV bracis-19054 179 2 , , PUNCT bracis-19054 179 3 the the DET bracis-19054 179 4 edge edge NOUN bracis-19054 179 5 weight weight NOUN bracis-19054 179 6 for for ADP bracis-19054 179 7 an an DET bracis-19054 179 8 object object NOUN bracis-19054 179 9 \(o_{i \(o_{i NOUN bracis-19054 179 10 } } PUNCT bracis-19054 179 11 \in \in PROPN bracis-19054 179 12 \mathcal \mathcal PROPN bracis-19054 179 13 { { PUNCT bracis-19054 179 14 o}_{l}\ o}_{l}\ PROPN bracis-19054 179 15 ) ) PUNCT bracis-19054 179 16 and and CCONJ bracis-19054 179 17 and and CCONJ bracis-19054 179 18 object object VERB bracis-19054 179 19 \(o_{j \(o_{j NOUN bracis-19054 179 20 } } PUNCT bracis-19054 179 21 \in \in PROPN bracis-19054 179 22 \mathcal \mathcal PROPN bracis-19054 179 23 { { PUNCT bracis-19054 179 24 o}_m\ o}_m\ NOUN bracis-19054 179 25 ) ) PUNCT bracis-19054 179 26 is be AUX bracis-19054 179 27 given give VERB bracis-19054 179 28 by by ADP bracis-19054 179 29 eq eq PROPN bracis-19054 179 30 . . PUNCT bracis-19054 179 31     SPACE bracis-19054 180 1 2 2 NUM bracis-19054 180 2 : : PUNCT bracis-19054 180 3 $ $ SYM bracis-19054 180 4 $ $ SYM bracis-19054 180 5 \begin{aligned \begin{aligne VERB bracis-19054 180 6 } } PUNCT bracis-19054 180 7 w_{o_i w_{o_i NOUN bracis-19054 180 8 , , PUNCT bracis-19054 180 9 o_j o_j SYM bracis-19054 180 10 } } PUNCT bracis-19054 180 11 = = SYM bracis-19054 180 12 \frac{w_{o_i \frac{w_{o_i NOUN bracis-19054 180 13 , , PUNCT bracis-19054 180 14 o_j}}{\sum o_j}}{\sum NOUN bracis-19054 180 15 _ _ PUNCT bracis-19054 180 16 { { PUNCT bracis-19054 180 17 o_{k o_{k PROPN bracis-19054 180 18 } } PUNCT bracis-19054 180 19 } } PUNCT bracis-19054 180 20 w_{o_i w_{o_i NOUN bracis-19054 180 21 , , PUNCT bracis-19054 180 22 o_k o_k NOUN bracis-19054 180 23 } } PUNCT bracis-19054 180 24 } } PUNCT bracis-19054 180 25 , , PUNCT bracis-19054 180 26 o_{i o_{i ADJ bracis-19054 180 27 } } PUNCT bracis-19054 180 28 \in \in PROPN bracis-19054 180 29 \mathcal \mathcal PROPN bracis-19054 180 30 { { PUNCT bracis-19054 180 31 o}_{l o}_{l NOUN bracis-19054 180 32 } } PUNCT bracis-19054 180 33 , , PUNCT bracis-19054 180 34 { { PUNCT bracis-19054 180 35 o_{j o_{j NOUN bracis-19054 180 36 } } PUNCT bracis-19054 180 37 \in \in PROPN bracis-19054 180 38 \mathcal \mathcal PROPN bracis-19054 180 39 { { PUNCT bracis-19054 180 40 o}_m o}_m NUM bracis-19054 180 41 } } PUNCT bracis-19054 180 42 , , PUNCT bracis-19054 180 43 \text \text NOUN bracis-19054 180 44 { { PUNCT bracis-19054 180 45 and and CCONJ bracis-19054 180 46 } } PUNCT bracis-19054 180 47 o_{k o_{k PROPN bracis-19054 180 48 } } PUNCT bracis-19054 180 49 \in \in PROPN bracis-19054 180 50 \mathcal \mathcal PROPN bracis-19054 180 51 { { PUNCT bracis-19054 180 52 o}_{m o}_{m NUM bracis-19054 180 53 } } PUNCT bracis-19054 180 54 \end{aligned}$$ \end{aligned}$$ X bracis-19054 180 55 ( ( PUNCT bracis-19054 180 56 2 2 NUM bracis-19054 180 57 ) ) PUNCT bracis-19054 180 58 after after ADP bracis-19054 180 59 building build VERB bracis-19054 180 60 the the DET bracis-19054 180 61 news news NOUN bracis-19054 180 62 and and CCONJ bracis-19054 180 63 terms term NOUN bracis-19054 180 64 network network NOUN bracis-19054 180 65 , , PUNCT bracis-19054 180 66 the the DET bracis-19054 180 67 next next ADJ bracis-19054 180 68 stage stage NOUN bracis-19054 180 69 ( ( PUNCT bracis-19054 180 70 fig fig NOUN bracis-19054 180 71 . . PUNCT bracis-19054 180 72     SPACE bracis-19054 180 73 1 1 NUM bracis-19054 180 74 stage stage NOUN bracis-19054 180 75 9 9 NUM bracis-19054 180 76 ) ) PUNCT bracis-19054 180 77 is be AUX bracis-19054 180 78 to to PART bracis-19054 180 79 carry carry VERB bracis-19054 180 80 out out ADP bracis-19054 180 81 the the DET bracis-19054 180 82 label label NOUN bracis-19054 180 83 propagation propagation NOUN bracis-19054 180 84 using use VERB bracis-19054 180 85 transductive transductive ADJ bracis-19054 180 86 learning learn VERB bracis-19054 180 87 algorithms algorithm NOUN bracis-19054 180 88 for for ADP bracis-19054 180 89 heterogeneous heterogeneous ADJ bracis-19054 180 90 networks network NOUN bracis-19054 180 91 . . PUNCT bracis-19054 181 1 4.4 4.4 NUM bracis-19054 181 2 label label NOUN bracis-19054 181 3 propagation propagation NOUN bracis-19054 181 4 we we PRON bracis-19054 181 5 propose propose VERB bracis-19054 181 6 using use VERB bracis-19054 181 7 regularization regularization NOUN bracis-19054 181 8 - - PUNCT bracis-19054 181 9 based base VERB bracis-19054 181 10 transductive transductive ADJ bracis-19054 181 11 algorithms algorithm NOUN bracis-19054 181 12 for for ADP bracis-19054 181 13 heterogeneous heterogeneous ADJ bracis-19054 181 14 networks network NOUN bracis-19054 181 15 to to PART bracis-19054 181 16 classify classify VERB bracis-19054 181 17 the the DET bracis-19054 181 18 unlabeled unlabeled ADJ bracis-19054 181 19 news news NOUN bracis-19054 181 20 as as ADP bracis-19054 181 21 fake fake ADJ bracis-19054 181 22 or or CCONJ bracis-19054 181 23 real real ADJ bracis-19054 181 24 . . PUNCT bracis-19054 182 1 the the DET bracis-19054 182 2 regularization regularization NOUN bracis-19054 182 3 - - PUNCT bracis-19054 182 4 based base VERB bracis-19054 182 5 algorithms algorithm NOUN bracis-19054 182 6 satisfy satisfy VERB bracis-19054 182 7 two two NUM bracis-19054 182 8 premises premise NOUN bracis-19054 182 9 : : PUNCT bracis-19054 182 10 ( ( PUNCT bracis-19054 182 11 i i NOUN bracis-19054 182 12 ) ) PUNCT bracis-19054 182 13 the the DET bracis-19054 182 14 class class NOUN bracis-19054 182 15 information information NOUN bracis-19054 182 16 of of ADP bracis-19054 182 17 neighboring neighboring NOUN bracis-19054 182 18 objects object NOUN bracis-19054 182 19 must must AUX bracis-19054 182 20 be be AUX bracis-19054 182 21 similar similar ADJ bracis-19054 182 22 ; ; PUNCT bracis-19054 182 23 ( ( PUNCT bracis-19054 182 24 ii ii NOUN bracis-19054 182 25 ) ) PUNCT bracis-19054 182 26 and and CCONJ bracis-19054 182 27 the the DET bracis-19054 182 28 class class NOUN bracis-19054 182 29 information information NOUN bracis-19054 182 30 of of ADP bracis-19054 182 31 the the DET bracis-19054 182 32 labeled label VERB bracis-19054 182 33 objects object NOUN bracis-19054 182 34 assigned assign VERB bracis-19054 182 35 during during ADP bracis-19054 182 36 the the DET bracis-19054 182 37 classification classification NOUN bracis-19054 182 38 process process NOUN bracis-19054 182 39 must must AUX bracis-19054 182 40 be be AUX bracis-19054 182 41 similar similar ADJ bracis-19054 182 42 to to ADP bracis-19054 182 43 the the DET bracis-19054 182 44 real real ADJ bracis-19054 182 45 class class NOUN bracis-19054 182 46 information information NOUN bracis-19054 182 47 [ [ X bracis-19054 182 48 12 12 NUM bracis-19054 182 49 ] ] PUNCT bracis-19054 182 50 . . PUNCT bracis-19054 183 1 the the DET bracis-19054 183 2 two two NUM bracis-19054 183 3 algorithms algorithm NOUN bracis-19054 183 4 considered consider VERB bracis-19054 183 5 in in ADP bracis-19054 183 6 this this DET bracis-19054 183 7 paper paper NOUN bracis-19054 183 8 are be AUX bracis-19054 183 9 : : PUNCT bracis-19054 183 10 label label NOUN bracis-19054 183 11 propagation propagation NOUN bracis-19054 183 12 through through ADP bracis-19054 183 13 heterogeneous heterogeneous ADJ bracis-19054 183 14 networks network NOUN bracis-19054 183 15 ( ( PUNCT bracis-19054 183 16 lphn lphn PROPN bracis-19054 183 17 ) ) PUNCT bracis-19054 184 1 [ [ X bracis-19054 184 2 12 12 NUM bracis-19054 184 3 ] ] PUNCT bracis-19054 184 4 and and CCONJ bracis-19054 184 5 gnetmine gnetmine PROPN bracis-19054 184 6 ( ( PUNCT bracis-19054 184 7 gnm gnm PROPN bracis-19054 184 8 ) ) PUNCT bracis-19054 185 1 [ [ X bracis-19054 185 2 7 7 NUM bracis-19054 185 3 ] ] PUNCT bracis-19054 185 4 . . PUNCT bracis-19054 186 1 in in ADP bracis-19054 186 2 order order NOUN bracis-19054 186 3 to to PART bracis-19054 186 4 explain explain VERB bracis-19054 186 5 the the DET bracis-19054 186 6 regularization regularization NOUN bracis-19054 186 7 functions function NOUN bracis-19054 186 8 of of ADP bracis-19054 186 9 both both DET bracis-19054 186 10 algorithms algorithm NOUN bracis-19054 186 11 , , PUNCT bracis-19054 186 12 let let VERB bracis-19054 186 13 \(\mathbf \(\mathbf VERB bracis-19054 186 14 { { PUNCT bracis-19054 186 15 f}_{o_i f}_{o_i NOUN bracis-19054 186 16 } } PUNCT bracis-19054 186 17 = = SYM bracis-19054 186 18 \{f_{\textit{interest \{f_{\textit{interest PROPN bracis-19054 186 19 } } PUNCT bracis-19054 186 20 } } PUNCT bracis-19054 186 21 , , PUNCT bracis-19054 186 22 f_{\textit{not f_{\textit{not NOUN bracis-19054 186 23 - - PUNCT bracis-19054 186 24 interest}}\}\ interest}}\}\ NOUN bracis-19054 186 25 ) ) PUNCT bracis-19054 186 26 be be VERB bracis-19054 186 27 the the DET bracis-19054 186 28 class class NOUN bracis-19054 186 29 information information NOUN bracis-19054 186 30 vector vector NOUN bracis-19054 186 31 , , PUNCT bracis-19054 186 32 which which PRON bracis-19054 186 33 gives give VERB bracis-19054 186 34 how how SCONJ bracis-19054 186 35 much much ADJ bracis-19054 186 36 an an DET bracis-19054 186 37 object object NOUN bracis-19054 186 38 \(o_i\ \(o_i\ NOUN bracis-19054 186 39 ) ) PUNCT bracis-19054 186 40 belongs belong VERB bracis-19054 186 41 to to ADP bracis-19054 186 42 each each PRON bracis-19054 186 43 of of ADP bracis-19054 186 44 the the DET bracis-19054 186 45 classes class NOUN bracis-19054 186 46 , , PUNCT bracis-19054 186 47 and and CCONJ bracis-19054 186 48 let let VERB bracis-19054 186 49 \(\mathbf \(\mathbf VERB bracis-19054 186 50 { { PUNCT bracis-19054 186 51 y}_{o_i}\ y}_{o_i}\ NOUN bracis-19054 186 52 ) ) PUNCT bracis-19054 186 53 be be AUX bracis-19054 186 54 the the DET bracis-19054 186 55 real real ADJ bracis-19054 186 56 class class NOUN bracis-19054 186 57 information information NOUN bracis-19054 186 58 vector vector NOUN bracis-19054 186 59 \(\mathbf \(\mathbf X bracis-19054 186 60 { { PUNCT bracis-19054 186 61 f}\ f}\ NOUN bracis-19054 186 62 ) ) PUNCT bracis-19054 186 63 , , PUNCT bracis-19054 186 64 in in ADP bracis-19054 186 65 which which PRON bracis-19054 186 66 the the DET bracis-19054 186 67 position position NOUN bracis-19054 186 68 of of ADP bracis-19054 186 69 the the DET bracis-19054 186 70 vector vector NOUN bracis-19054 186 71 corresponding correspond VERB bracis-19054 186 72 to to ADP bracis-19054 186 73 its its PRON bracis-19054 186 74 class class NOUN bracis-19054 186 75 is be AUX bracis-19054 186 76 filled fill VERB bracis-19054 186 77 with with ADP bracis-19054 186 78 1 1 NUM bracis-19054 186 79 . . PUNCT bracis-19054 187 1 thus thus ADV bracis-19054 187 2 , , PUNCT bracis-19054 187 3 only only ADV bracis-19054 187 4 labeled label VERB bracis-19054 187 5 objects object NOUN bracis-19054 187 6 have have VERB bracis-19054 187 7 values value NOUN bracis-19054 187 8 for for ADP bracis-19054 187 9 vector vector NOUN bracis-19054 187 10 \(\mathbf \(\mathbf X bracis-19054 187 11 { { PUNCT bracis-19054 187 12 y}\ y}\ NOUN bracis-19054 187 13 ) ) PUNCT bracis-19054 187 14 . . PUNCT bracis-19054 188 1 the the DET bracis-19054 188 2 term term NOUN bracis-19054 188 3 \ \ PROPN bracis-19054 188 4 ( ( PUNCT bracis-19054 188 5 w_{o_i w_{o_i NOUN bracis-19054 188 6 , , PUNCT bracis-19054 188 7 o_j}\ o_j}\ PROPN bracis-19054 188 8 ) ) PUNCT bracis-19054 188 9 is be AUX bracis-19054 188 10 the the DET bracis-19054 188 11 weight weight NOUN bracis-19054 188 12 of of ADP bracis-19054 188 13 the the DET bracis-19054 188 14 edge edge NOUN bracis-19054 188 15 connecting connect VERB bracis-19054 188 16 object object NOUN bracis-19054 188 17 \(o_i\ \(o_i\ NOUN bracis-19054 188 18 ) ) PUNCT bracis-19054 188 19 to to PART bracis-19054 188 20 object object VERB bracis-19054 188 21 \(o_j\ \(o_j\ PROPN bracis-19054 188 22 ) ) PUNCT bracis-19054 188 23 , , PUNCT bracis-19054 188 24 and and CCONJ bracis-19054 188 25 \(\mathcal \(\mathcal ADJ bracis-19054 188 26 { { PUNCT bracis-19054 188 27 o}^{l}\ o}^{l}\ NOUN bracis-19054 188 28 ) ) PUNCT bracis-19054 188 29 refers refer VERB bracis-19054 188 30 to to ADP bracis-19054 188 31 the the DET bracis-19054 188 32 set set NOUN bracis-19054 188 33 of of ADP bracis-19054 188 34 labeled label VERB bracis-19054 188 35 objects object NOUN bracis-19054 188 36 . . PUNCT bracis-19054 189 1 the the DET bracis-19054 189 2 regularization regularization NOUN bracis-19054 189 3 function function NOUN bracis-19054 189 4 to to PART bracis-19054 189 5 be be AUX bracis-19054 189 6 minimized minimize VERB bracis-19054 189 7 by by ADP bracis-19054 189 8 lphn lphn PROPN bracis-19054 189 9 is be AUX bracis-19054 189 10 given give VERB bracis-19054 189 11 by by ADP bracis-19054 189 12 : : PUNCT bracis-19054 189 13 $ $ SYM bracis-19054 189 14 $ $ SYM bracis-19054 189 15 \begin{aligned \begin{aligne VERB bracis-19054 189 16 } } PUNCT bracis-19054 189 17 q(\mathbf q(\mathbf VERB bracis-19054 189 18 { { PUNCT bracis-19054 189 19 f f NOUN bracis-19054 189 20 } } PUNCT bracis-19054 189 21 ) ) PUNCT bracis-19054 190 1 = = SYM bracis-19054 190 2 \sum \sum NOUN bracis-19054 190 3 _ _ PUNCT bracis-19054 190 4 { { PUNCT bracis-19054 190 5 \mathcal \mathcal ADJ bracis-19054 190 6 { { PUNCT bracis-19054 190 7 d},\mathcal d},\mathcal ADJ bracis-19054 190 8 { { PUNCT bracis-19054 190 9 t t PROPN bracis-19054 190 10 } } PUNCT bracis-19054 190 11 \in \in PROPN bracis-19054 190 12 \mathcal \mathcal PROPN bracis-19054 190 13 { { PUNCT bracis-19054 190 14 o o NOUN bracis-19054 190 15 } } PUNCT bracis-19054 190 16 } } PUNCT bracis-19054 190 17 \frac{1}{2 \frac{1}{2 NOUN bracis-19054 190 18 } } PUNCT bracis-19054 190 19 \sum \sum NOUN bracis-19054 190 20 _ _ PUNCT bracis-19054 190 21 { { PUNCT bracis-19054 190 22 { { PUNCT bracis-19054 190 23 o_i o_i NOUN bracis-19054 190 24 } } PUNCT bracis-19054 190 25 \in \in PROPN bracis-19054 190 26 \mathcal \mathcal PROPN bracis-19054 190 27 { { PUNCT bracis-19054 190 28 o}_i o}_i ADJ bracis-19054 190 29 } } PUNCT bracis-19054 190 30 \sum \sum NOUN bracis-19054 190 31 _ _ PUNCT bracis-19054 190 32 { { PUNCT bracis-19054 190 33 { { PUNCT bracis-19054 190 34 o_j o_j PUNCT bracis-19054 190 35 } } SYM bracis-19054 190 36 \in \in PROPN bracis-19054 190 37 \mathcal \mathcal PROPN bracis-19054 190 38 { { PUNCT bracis-19054 190 39 o}_j o}_j NOUN bracis-19054 190 40 } } PUNCT bracis-19054 190 41 w_{o_i w_{o_i NOUN bracis-19054 190 42 , , PUNCT bracis-19054 190 43 o_j}(\mathbf o_j}(\mathbf NUM bracis-19054 190 44 { { PUNCT bracis-19054 190 45 f}_{o_i f}_{o_i NOUN bracis-19054 190 46 } } PUNCT bracis-19054 190 47 \mathbf \mathbf PROPN bracis-19054 190 48 { { PUNCT bracis-19054 190 49 f}_{o_j})^2 f}_{o_j})^2 ADJ bracis-19054 190 50 + + CCONJ bracis-19054 190 51 \lim \lim NOUN bracis-19054 190 52 _ _ PUNCT bracis-19054 190 53 { { PUNCT bracis-19054 190 54 \mu \mu NOUN bracis-19054 190 55 \rightarrow \rightarrow PROPN bracis-19054 190 56 \infty \infty PROPN bracis-19054 190 57 } } PUNCT bracis-19054 190 58 \sum \sum NOUN bracis-19054 190 59 _ _ NOUN bracis-19054 190 60 { { PUNCT bracis-19054 190 61 o_i o_i NUM bracis-19054 190 62 \in \in PROPN bracis-19054 190 63 \mathcal \mathcal PROPN bracis-19054 190 64 { { PUNCT bracis-19054 190 65 o}^l o}^l NOUN bracis-19054 190 66 } } PUNCT bracis-19054 190 67 ( ( PUNCT bracis-19054 190 68 \mathbf \mathbf PROPN bracis-19054 190 69 { { PUNCT bracis-19054 190 70 f}_{o_i f}_{o_i NOUN bracis-19054 190 71 } } PUNCT bracis-19054 190 72 \mathbf \mathbf PROPN bracis-19054 190 73 { { PUNCT bracis-19054 190 74 y}_{o_i})^2 y}_{o_i})^2 PROPN bracis-19054 190 75 . . PUNCT bracis-19054 191 1 \end{aligned}$$ \end{aligned}$$ X bracis-19054 191 2 ( ( PUNCT bracis-19054 191 3 3 3 X bracis-19054 191 4 ) ) PUNCT bracis-19054 191 5 we we PRON bracis-19054 191 6 can can AUX bracis-19054 191 7 observe observe VERB bracis-19054 191 8 that that SCONJ bracis-19054 191 9 the the DET bracis-19054 191 10 different different ADJ bracis-19054 191 11 relations relation NOUN bracis-19054 191 12 have have VERB bracis-19054 191 13 the the DET bracis-19054 191 14 same same ADJ bracis-19054 191 15 importance importance NOUN bracis-19054 191 16 and and CCONJ bracis-19054 191 17 that that SCONJ bracis-19054 191 18 \(\lim \(\lim NOUN bracis-19054 191 19 _ _ PUNCT bracis-19054 191 20 { { PUNCT bracis-19054 191 21 \mu \mu AUX bracis-19054 191 22 \rightarrow \rightarrow PROPN bracis-19054 191 23 \infty \infty PROPN bracis-19054 191 24 } } PUNCT bracis-19054 191 25 \ \ PROPN bracis-19054 191 26 ) ) PUNCT bracis-19054 191 27 , , PUNCT bracis-19054 191 28 which which PRON bracis-19054 191 29 forces force VERB bracis-19054 191 30 that that PRON bracis-19054 191 31 \(\mathbf \(\mathbf VERB bracis-19054 191 32 { { PUNCT bracis-19054 191 33 f}_{i f}_{i INTJ bracis-19054 191 34 } } PUNCT bracis-19054 191 35 = = SYM bracis-19054 191 36 \mathbf \mathbf PROPN bracis-19054 191 37 { { PUNCT bracis-19054 191 38 y}_{i},\ y}_{i},\ PROPN bracis-19054 191 39 , , PUNCT bracis-19054 191 40 \forall \forall DET bracis-19054 191 41 o_{i o_{i PROPN bracis-19054 191 42 } } PUNCT bracis-19054 191 43 \in \in PROPN bracis-19054 191 44 \mathcal \mathcal PROPN bracis-19054 191 45 { { PUNCT bracis-19054 191 46 o}^{l}\ o}^{l}\ NOUN bracis-19054 191 47 ) ) PUNCT bracis-19054 191 48 , , PUNCT bracis-19054 191 49 i.e. i.e. X bracis-19054 191 50 , , PUNCT bracis-19054 191 51 the the DET bracis-19054 191 52 class class NOUN bracis-19054 191 53 information information NOUN bracis-19054 191 54 of of ADP bracis-19054 191 55 label label NOUN bracis-19054 191 56 object object NOUN bracis-19054 191 57 do do AUX bracis-19054 191 58 not not PART bracis-19054 191 59 change change VERB bracis-19054 191 60 . . PUNCT bracis-19054 192 1 after after ADP bracis-19054 192 2 inferring infer VERB bracis-19054 192 3 the the DET bracis-19054 192 4 \(\mathbf \(\mathbf VERB bracis-19054 192 5 { { PUNCT bracis-19054 192 6 f}\ f}\ NOUN bracis-19054 192 7 ) ) PUNCT bracis-19054 192 8 vectors vector NOUN bracis-19054 192 9 of of ADP bracis-19054 192 10 the the DET bracis-19054 192 11 unlabeled unlabele VERB bracis-19054 192 12 objects object NOUN bracis-19054 192 13 , , PUNCT bracis-19054 192 14 the the DET bracis-19054 192 15 class class NOUN bracis-19054 192 16 mass mass NOUN bracis-19054 192 17 normalization normalization NOUN bracis-19054 192 18 ( ( PUNCT bracis-19054 192 19 cmn cmn NOUN bracis-19054 192 20 ) ) PUNCT bracis-19054 193 1 [ [ X bracis-19054 193 2 23 23 NUM bracis-19054 193 3 ] ] PUNCT bracis-19054 193 4 , , PUNCT bracis-19054 193 5 and and CCONJ bracis-19054 193 6 then then ADV bracis-19054 193 7 the the DET bracis-19054 193 8 objects object NOUN bracis-19054 193 9 are be AUX bracis-19054 193 10 classified classify VERB bracis-19054 193 11 according accord VERB bracis-19054 193 12 to to ADP bracis-19054 193 13 the the DET bracis-19054 193 14 \(\arg \(\arg NOUN bracis-19054 193 15 \)-\(\max \)-\(\max PROPN bracis-19054 193 16 \ \ PROPN bracis-19054 193 17 ) ) PUNCT bracis-19054 193 18 of of ADP bracis-19054 193 19 the the DET bracis-19054 193 20 corresponding corresponding ADJ bracis-19054 193 21 \(\mathbf \(\mathbf VERB bracis-19054 193 22 { { PUNCT bracis-19054 193 23 f}\ f}\ NOUN bracis-19054 193 24 ) ) PUNCT bracis-19054 193 25 vector vector NOUN bracis-19054 193 26 . . PUNCT bracis-19054 194 1 in in ADP bracis-19054 194 2 gnetmine gnetmine PROPN bracis-19054 194 3     SPACE bracis-19054 194 4 [ [ X bracis-19054 194 5 7 7 NUM bracis-19054 194 6 ] ] PUNCT bracis-19054 194 7 , , PUNCT bracis-19054 194 8 the the DET bracis-19054 194 9 relations relation NOUN bracis-19054 194 10 between between ADP bracis-19054 194 11 different different ADJ bracis-19054 194 12 objects object NOUN bracis-19054 194 13 have have VERB bracis-19054 194 14 different different ADJ bracis-19054 194 15 importance importance NOUN bracis-19054 194 16 . . PUNCT bracis-19054 195 1 the the DET bracis-19054 195 2 label label NOUN bracis-19054 195 3 of of ADP bracis-19054 195 4 an an DET bracis-19054 195 5 object object NOUN bracis-19054 195 6 \(o_{i \(o_{i NOUN bracis-19054 195 7 } } PUNCT bracis-19054 195 8 \in \in PROPN bracis-19054 195 9 \mathcal \mathcal PROPN bracis-19054 195 10 { { PUNCT bracis-19054 195 11 o}^{l}\ o}^{l}\ NOUN bracis-19054 195 12 ) ) PUNCT bracis-19054 195 13 can can AUX bracis-19054 195 14 be be AUX bracis-19054 195 15 changed change VERB bracis-19054 195 16 during during ADP bracis-19054 195 17 the the DET bracis-19054 195 18 classification classification NOUN bracis-19054 195 19 process process NOUN bracis-19054 195 20 if if SCONJ bracis-19054 195 21 information information NOUN bracis-19054 195 22 from from ADP bracis-19054 195 23 neighboring neighboring NOUN bracis-19054 195 24 objects object NOUN bracis-19054 195 25 diverges diverge VERB bracis-19054 195 26 from from ADP bracis-19054 195 27 the the DET bracis-19054 195 28 class class NOUN bracis-19054 195 29 of of ADP bracis-19054 195 30 the the DET bracis-19054 195 31 object object NOUN bracis-19054 195 32 initially initially ADV bracis-19054 195 33 labeled label VERB bracis-19054 195 34 . . PUNCT bracis-19054 196 1 the the DET bracis-19054 196 2 regularization regularization NOUN bracis-19054 196 3 performed perform VERB bracis-19054 196 4 by by ADP bracis-19054 196 5 gnetmine gnetmine PROPN bracis-19054 196 6 is be AUX bracis-19054 196 7 given give VERB bracis-19054 196 8 by by ADP bracis-19054 196 9 : : PUNCT bracis-19054 196 10 $ $ SYM bracis-19054 196 11 $ $ SYM bracis-19054 196 12 \begin{aligned \begin{aligne VERB bracis-19054 196 13 } } PUNCT bracis-19054 196 14 q(\mathbf q(\mathbf VERB bracis-19054 196 15 { { PUNCT bracis-19054 196 16 f f NOUN bracis-19054 196 17 } } PUNCT bracis-19054 196 18 ) ) PUNCT bracis-19054 197 1 = = PROPN bracis-19054 197 2 & & CCONJ bracis-19054 197 3 \sum \sum NOUN bracis-19054 197 4 _ _ PUNCT bracis-19054 197 5 { { PUNCT bracis-19054 197 6 \mathcal \mathcal ADJ bracis-19054 197 7 { { PUNCT bracis-19054 197 8 o}_i,\mathcal o}_i,\mathcal ADJ bracis-19054 197 9 { { PUNCT bracis-19054 197 10 o}_j\subset o}_j\subset NOUN bracis-19054 197 11 \mathcal \mathcal X bracis-19054 197 12 { { PUNCT bracis-19054 197 13 o o NOUN bracis-19054 197 14 } } PUNCT bracis-19054 197 15 } } PUNCT bracis-19054 197 16 \lambda \lambda PROPN bracis-19054 197 17 _ _ PUNCT bracis-19054 197 18 { { PUNCT bracis-19054 197 19 \mathcal \mathcal ADJ bracis-19054 197 20 { { PUNCT bracis-19054 197 21 o}_i,\mathcal o}_i,\mathcal ADJ bracis-19054 197 22 { { PUNCT bracis-19054 197 23 o}_j o}_j NOUN bracis-19054 197 24 } } PUNCT bracis-19054 197 25 \sum \sum NOUN bracis-19054 197 26 _ _ PUNCT bracis-19054 197 27 { { PUNCT bracis-19054 197 28 o_k\in o_k\in PROPN bracis-19054 197 29 \mathcal \mathcal PROPN bracis-19054 197 30 { { PUNCT bracis-19054 197 31 o}_i o}_i ADJ bracis-19054 197 32 } } PUNCT bracis-19054 197 33 \sum \sum NOUN bracis-19054 197 34 _ _ NOUN bracis-19054 197 35 { { PUNCT bracis-19054 197 36 o_l\in o_l\in ADP bracis-19054 197 37 \mathcal \mathcal PROPN bracis-19054 197 38 { { PUNCT bracis-19054 197 39 o}j o}j NOUN bracis-19054 197 40 } } PUNCT bracis-19054 197 41 w_{o_k w_{o_k NOUN bracis-19054 197 42 , , PUNCT bracis-19054 197 43 o_l o_l SYM bracis-19054 197 44 } } PUNCT bracis-19054 197 45 \left| \left| NOUN bracis-19054 197 46 \left| \left| NOUN bracis-19054 197 47 \frac{\mathbf \frac{\mathbf ADJ bracis-19054 197 48 { { PUNCT bracis-19054 197 49 f}_{o_k}(\mathcal f}_{o_k}(\mathcal ADJ bracis-19054 197 50 { { PUNCT bracis-19054 197 51 o}_i)}{\sqrt{\displaystyle o}_i)}{\sqrt{\displaystyle PROPN bracis-19054 197 52 \sum \sum NOUN bracis-19054 197 53 _ _ PUNCT bracis-19054 197 54 { { PUNCT bracis-19054 197 55 o_m\in o_m\in PROPN bracis-19054 197 56 \mathcal \mathcal PROPN bracis-19054 197 57 { { PUNCT bracis-19054 197 58 o}_j}w_{o_k o}_j}w_{o_k ADJ bracis-19054 197 59 , , PUNCT bracis-19054 197 60 o_m o_m ADV bracis-19054 197 61 } } PUNCT bracis-19054 197 62 } } PUNCT bracis-19054 197 63 } } PUNCT bracis-19054 197 64 \frac{\mathbf \frac{\mathbf PROPN bracis-19054 197 65 { { PUNCT bracis-19054 197 66 f}_{o_l}(\mathcal f}_{o_l}(\mathcal ADJ bracis-19054 197 67 { { PUNCT bracis-19054 197 68 o}_j)}{\sqrt{\displaystyle o}_j)}{\sqrt{\displaystyle PROPN bracis-19054 197 69 \sum \sum NOUN bracis-19054 197 70 _ _ PUNCT bracis-19054 197 71 { { PUNCT bracis-19054 197 72 o_m\in o_m\in PROPN bracis-19054 197 73 \mathcal \mathcal ADJ bracis-19054 197 74 { { PUNCT bracis-19054 197 75 o}_i}w_{o_l o}_i}w_{o_l NOUN bracis-19054 197 76 , , PUNCT bracis-19054 197 77 o_m o_m ADV bracis-19054 197 78 } } PUNCT bracis-19054 197 79 } } PUNCT bracis-19054 197 80 } } PUNCT bracis-19054 197 81 \right| \right| X bracis-19054 197 82 \right| \right| NOUN bracis-19054 197 83 ^2 ^2 PUNCT bracis-19054 197 84 \nonumber \nonumber PROPN bracis-19054 197 85 \\&\quad \\&\quad PROPN bracis-19054 197 86 \quad \quad VERB bracis-19054 197 87 \qquad \qquad PROPN bracis-19054 197 88 \qquad \qquad PROPN bracis-19054 197 89 \qquad \qquad PROPN bracis-19054 197 90 \qquad \qquad PROPN bracis-19054 197 91 \qquad \qquad PROPN bracis-19054 197 92 \qquad \qquad PROPN bracis-19054 197 93 \quad \quad PROPN bracis-19054 197 94 \qquad \qquad PROPN bracis-19054 197 95 \;\ \;\ NOUN bracis-19054 197 96 ; ; PUNCT bracis-19054 197 97 + + X bracis-19054 197 98 \sum \sum NOUN bracis-19054 197 99 _ _ PUNCT bracis-19054 197 100 { { PUNCT bracis-19054 197 101 o_j\in o_j\in NOUN bracis-19054 197 102 \mathcal \mathcal PROPN bracis-19054 197 103 { { PUNCT bracis-19054 197 104 o}^l}\alpha o}^l}\alpha NOUN bracis-19054 197 105 _ _ PRON bracis-19054 197 106 { { PUNCT bracis-19054 197 107 o_j}(\mathbf o_j}(\mathbf NOUN bracis-19054 197 108 { { PUNCT bracis-19054 197 109 f}{o_j}-\mathbf f}{o_j}-\mathbf PROPN bracis-19054 197 110 { { PUNCT bracis-19054 197 111 y}{o_j y}{o_j PROPN bracis-19054 197 112 } } PUNCT bracis-19054 197 113 ) ) PUNCT bracis-19054 197 114 \end{aligned}$$ \end{aligned}$$ PROPN bracis-19054 197 115 ( ( PUNCT bracis-19054 197 116 4 4 NUM bracis-19054 197 117 ) ) PUNCT bracis-19054 197 118 in in ADP bracis-19054 197 119 which which PRON bracis-19054 197 120 \(\lambda \(\lambda NOUN bracis-19054 197 121 _ _ PUNCT bracis-19054 197 122 { { PUNCT bracis-19054 197 123 \mathcal \mathcal ADJ bracis-19054 197 124 { { PUNCT bracis-19054 197 125 o}_i,\mathcal o}_i,\mathcal ADJ bracis-19054 197 126 { { PUNCT bracis-19054 197 127 o}_j o}_j NOUN bracis-19054 197 128 } } PUNCT bracis-19054 197 129 ( ( PUNCT bracis-19054 197 130 0 0 NUM bracis-19054 197 131 \ge \ge PROPN bracis-19054 197 132 \lambda \lambda PROPN bracis-19054 197 133 _ _ PUNCT bracis-19054 197 134 { { PUNCT bracis-19054 197 135 \mathcal \mathcal ADJ bracis-19054 197 136 { { PUNCT bracis-19054 197 137 o}_i,\mathcal o}_i,\mathcal ADJ bracis-19054 197 138 { { PUNCT bracis-19054 197 139 o}_j o}_j NOUN bracis-19054 197 140 } } PUNCT bracis-19054 197 141 \ge \ge PROPN bracis-19054 197 142 1)\ 1)\ NUM bracis-19054 197 143 ) ) PUNCT bracis-19054 197 144 is be AUX bracis-19054 197 145 the the DET bracis-19054 197 146 importance importance NOUN bracis-19054 197 147 given give VERB bracis-19054 197 148 to to ADP bracis-19054 197 149 the the DET bracis-19054 197 150 relationship relationship NOUN bracis-19054 197 151 between between ADP bracis-19054 197 152 objects object NOUN bracis-19054 197 153 of of ADP bracis-19054 197 154 the the DET bracis-19054 197 155 types type NOUN bracis-19054 197 156 \(\mathcal \(\mathcal ADJ bracis-19054 197 157 { { PUNCT bracis-19054 197 158 o}_i\ o}_i\ NOUN bracis-19054 197 159 ) ) PUNCT bracis-19054 197 160 and and CCONJ bracis-19054 197 161 \(\mathcal \(\mathcal ADJ bracis-19054 197 162 { { PUNCT bracis-19054 197 163 o}_j\ o}_j\ NOUN bracis-19054 197 164 ) ) PUNCT bracis-19054 197 165 , , PUNCT bracis-19054 197 166 and and CCONJ bracis-19054 197 167 \(\alpha \(\alpha VERB bracis-19054 197 168 _ _ PRON bracis-19054 197 169 { { PUNCT bracis-19054 197 170 o_i o_i NOUN bracis-19054 197 171 } } PUNCT bracis-19054 197 172 ( ( PUNCT bracis-19054 197 173 0 0 NUM bracis-19054 197 174 \ge \ge PROPN bracis-19054 197 175 \alpha \alpha ADJ bracis-19054 197 176 _ _ NOUN bracis-19054 197 177 { { PUNCT bracis-19054 197 178 o_i o_i NOUN bracis-19054 197 179 } } PUNCT bracis-19054 197 180 \ge \ge PROPN bracis-19054 197 181 1)\ 1)\ NUM bracis-19054 197 182 ) ) PUNCT bracis-19054 197 183 is be AUX bracis-19054 197 184 the the DET bracis-19054 197 185 importance importance NOUN bracis-19054 197 186 given give VERB bracis-19054 197 187 to to ADP bracis-19054 197 188 the the DET bracis-19054 197 189 real real ADJ bracis-19054 197 190 class class NOUN bracis-19054 197 191 information information NOUN bracis-19054 197 192 of of ADP bracis-19054 197 193 an an DET bracis-19054 197 194 object object NOUN bracis-19054 197 195 \(o_j \(o_j NOUN bracis-19054 197 196 \in \in PROPN bracis-19054 197 197 \mathcal \mathcal PROPN bracis-19054 197 198 { { PUNCT bracis-19054 197 199 o}^{l}\ o}^{l}\ NOUN bracis-19054 197 200 ) ) PUNCT bracis-19054 197 201 ( ( PUNCT bracis-19054 197 202 set set VERB bracis-19054 197 203 of of ADP bracis-19054 197 204 labeled label VERB bracis-19054 197 205 objects object NOUN bracis-19054 197 206 ) ) PUNCT bracis-19054 197 207 . . PUNCT bracis-19054 198 1 the the DET bracis-19054 198 2 documents document NOUN bracis-19054 198 3 are be AUX bracis-19054 198 4 classified classify VERB bracis-19054 198 5 considering consider VERB bracis-19054 198 6 the the DET bracis-19054 198 7 arg arg NOUN bracis-19054 198 8 - - PUNCT bracis-19054 198 9 max max PROPN bracis-19054 198 10 of of ADP bracis-19054 198 11 the the DET bracis-19054 198 12 final final ADJ bracis-19054 198 13 value value NOUN bracis-19054 198 14 of of ADP bracis-19054 198 15 \(\mathbf \(\mathbf PROPN bracis-19054 198 16 { { PUNCT bracis-19054 198 17 f}_{d_{i}}\ f}_{d_{i}}\ NOUN bracis-19054 198 18 ) ) PUNCT bracis-19054 198 19 vectors vector NOUN bracis-19054 198 20 for for ADP bracis-19054 198 21 \(d_{i \(d_{i ADP bracis-19054 198 22 } } SYM bracis-19054 198 23 \in \in PROPN bracis-19054 198 24 \mathcal \mathcal PROPN bracis-19054 198 25 { { PUNCT bracis-19054 198 26 d}^{u}\ d}^{u}\ NOUN bracis-19054 198 27 ) ) PUNCT bracis-19054 198 28 . . PUNCT bracis-19054 199 1 5 5 NUM bracis-19054 199 2 experimental experimental ADJ bracis-19054 199 3 evaluation evaluation NOUN bracis-19054 199 4 in in ADP bracis-19054 199 5 this this DET bracis-19054 199 6 section section NOUN bracis-19054 199 7 , , PUNCT bracis-19054 199 8 we we PRON bracis-19054 199 9 present present VERB bracis-19054 199 10 the the DET bracis-19054 199 11 experimental experimental ADJ bracis-19054 199 12 configuration configuration NOUN bracis-19054 199 13 used use VERB bracis-19054 199 14 in in ADP bracis-19054 199 15 the the DET bracis-19054 199 16 experiments experiment NOUN bracis-19054 199 17 , , PUNCT bracis-19054 199 18 the the DET bracis-19054 199 19 datasets dataset NOUN bracis-19054 199 20 and and CCONJ bracis-19054 199 21 an an DET bracis-19054 199 22 analysis analysis NOUN bracis-19054 199 23 of of ADP bracis-19054 199 24 the the DET bracis-19054 199 25 results result NOUN bracis-19054 199 26 achieved achieve VERB bracis-19054 199 27 . . PUNCT bracis-19054 200 1 our our PRON bracis-19054 200 2 goal goal NOUN bracis-19054 200 3 is be AUX bracis-19054 200 4 to to PART bracis-19054 200 5 encourage encourage VERB bracis-19054 200 6 the the DET bracis-19054 200 7 study study NOUN bracis-19054 200 8 of of ADP bracis-19054 200 9 pul pul NOUN bracis-19054 200 10 approaches approach NOUN bracis-19054 200 11 to to ADP bracis-19054 200 12 fake fake ADJ bracis-19054 200 13 news news NOUN bracis-19054 200 14 detection detection NOUN bracis-19054 200 15 , , PUNCT bracis-19054 200 16 which which PRON bracis-19054 200 17 perform perform VERB bracis-19054 200 18 well well ADV bracis-19054 200 19 using use VERB bracis-19054 200 20 little little ADJ bracis-19054 200 21 labeled label VERB bracis-19054 200 22 data datum NOUN bracis-19054 200 23 . . PUNCT bracis-19054 201 1 furthermore furthermore ADV bracis-19054 201 2 , , PUNCT bracis-19054 201 3 we we PRON bracis-19054 201 4 want want VERB bracis-19054 201 5 to to PART bracis-19054 201 6 demonstrate demonstrate VERB bracis-19054 201 7 that that SCONJ bracis-19054 201 8 with with ADP bracis-19054 201 9 little little ADJ bracis-19054 201 10 labeled label VERB bracis-19054 201 11 fake fake ADJ bracis-19054 201 12 news news NOUN bracis-19054 201 13 data datum NOUN bracis-19054 201 14 , , PUNCT bracis-19054 201 15 structured structure VERB bracis-19054 201 16 in in ADP bracis-19054 201 17 a a DET bracis-19054 201 18 network network NOUN bracis-19054 201 19 , , PUNCT bracis-19054 201 20 it it PRON bracis-19054 201 21 is be AUX bracis-19054 201 22 possible possible ADJ bracis-19054 201 23 to to PART bracis-19054 201 24 achieve achieve VERB bracis-19054 201 25 reasonable reasonable ADJ bracis-19054 201 26 classification classification NOUN bracis-19054 201 27 performance performance NOUN bracis-19054 201 28 , , PUNCT bracis-19054 201 29 which which PRON bracis-19054 201 30 can can AUX bracis-19054 201 31 be be AUX bracis-19054 201 32 improved improve VERB bracis-19054 201 33 by by ADP bracis-19054 201 34 adding add VERB bracis-19054 201 35 extra extra ADJ bracis-19054 201 36 information information NOUN bracis-19054 201 37 , , PUNCT bracis-19054 201 38 taken take VERB bracis-19054 201 39 from from ADP bracis-19054 201 40 the the DET bracis-19054 201 41 news news NOUN bracis-19054 201 42 textual textual ADJ bracis-19054 201 43 content content NOUN bracis-19054 201 44 . . PUNCT bracis-19054 202 1 table table NOUN bracis-19054 202 2 3 3 NUM bracis-19054 202 3 . . PUNCT bracis-19054 202 4 detailed detailed ADJ bracis-19054 202 5 information information NOUN bracis-19054 202 6 about about ADP bracis-19054 202 7 news news NOUN bracis-19054 202 8 datasets.full datasets.full NUM bracis-19054 202 9 size size NOUN bracis-19054 202 10 table table NOUN bracis-19054 202 11 5.1 5.1 NUM bracis-19054 202 12 news news NOUN bracis-19054 202 13 datasets dataset NOUN bracis-19054 202 14 in in ADP bracis-19054 202 15 this this DET bracis-19054 202 16 paper paper NOUN bracis-19054 202 17 , , PUNCT bracis-19054 202 18 six six NUM bracis-19054 202 19 news news NOUN bracis-19054 202 20 datasetsfootnote datasetsfootnote NOUN bracis-19054 202 21 1 1 NUM bracis-19054 202 22 were be AUX bracis-19054 202 23 evaluated evaluate VERB bracis-19054 202 24 , , PUNCT bracis-19054 202 25 four four NUM bracis-19054 202 26 of of ADP bracis-19054 202 27 them they PRON bracis-19054 202 28 are be AUX bracis-19054 202 29 in in ADP bracis-19054 202 30 english english PROPN bracis-19054 202 31 and and CCONJ bracis-19054 202 32 two two NUM bracis-19054 202 33 are be AUX bracis-19054 202 34 in in ADP bracis-19054 202 35 portuguese portuguese ADJ bracis-19054 202 36 . . PUNCT bracis-19054 203 1 table table NOUN bracis-19054 203 2     SPACE bracis-19054 203 3 3 3 NUM bracis-19054 203 4 presents present VERB bracis-19054 203 5 detailed detailed ADJ bracis-19054 203 6 information information NOUN bracis-19054 203 7 about about ADP bracis-19054 203 8 language language NOUN bracis-19054 203 9 , , PUNCT bracis-19054 203 10 subject subject NOUN bracis-19054 203 11 and and CCONJ bracis-19054 203 12 amount amount NOUN bracis-19054 203 13 of of ADP bracis-19054 203 14 real real ADJ bracis-19054 203 15 and and CCONJ bracis-19054 203 16 fake fake ADJ bracis-19054 203 17 news news NOUN bracis-19054 203 18 present present NOUN bracis-19054 203 19 in in ADP bracis-19054 203 20 each each DET bracis-19054 203 21 dataset dataset NOUN bracis-19054 203 22 . . PUNCT bracis-19054 204 1 fact fact NOUN bracis-19054 204 2 - - PUNCT bracis-19054 204 3 checked check VERB bracis-19054 204 4 news news NOUN bracis-19054 204 5 ( ( PUNCT bracis-19054 204 6 fcn fcn PROPN bracis-19054 204 7 ) ) PUNCT bracis-19054 204 8 was be AUX bracis-19054 204 9 collected collect VERB bracis-19054 204 10 from from ADP bracis-19054 204 11 five five NUM bracis-19054 204 12 brazilian brazilian ADJ bracis-19054 204 13 journalistic journalistic ADJ bracis-19054 204 14 fact fact NOUN bracis-19054 204 15 - - PUNCT bracis-19054 204 16 checking check VERB bracis-19054 204 17 sites site NOUN bracis-19054 204 18 to to PART bracis-19054 204 19 evaluate evaluate VERB bracis-19054 204 20 our our PRON bracis-19054 204 21 approach approach NOUN bracis-19054 204 22 . . PUNCT bracis-19054 205 1 the the DET bracis-19054 205 2 second second ADJ bracis-19054 205 3 portuguese portuguese ADJ bracis-19054 205 4 dataset dataset NOUN bracis-19054 205 5 is be AUX bracis-19054 205 6 fake.br fake.br X bracis-19054 205 7 ( ( PUNCT bracis-19054 205 8 fbr fbr NOUN bracis-19054 205 9 ) ) PUNCT bracis-19054 205 10 , , PUNCT bracis-19054 205 11 the the DET bracis-19054 205 12 first first ADJ bracis-19054 205 13 reference reference NOUN bracis-19054 205 14 corpus corpus NOUN bracis-19054 205 15 in in ADP bracis-19054 205 16 portuguese portuguese PROPN bracis-19054 205 17 for for ADP bracis-19054 205 18 fake fake ADJ bracis-19054 205 19 news news NOUN bracis-19054 205 20 detection detection NOUN bracis-19054 205 21     SPACE bracis-19054 206 1 [ [ X bracis-19054 206 2 17 17 NUM bracis-19054 206 3 ] ] PUNCT bracis-19054 206 4 . . PUNCT bracis-19054 207 1 the the DET bracis-19054 207 2 third third ADJ bracis-19054 207 3 dataset dataset NOUN bracis-19054 207 4 was be AUX bracis-19054 207 5 acquired acquire VERB bracis-19054 207 6 from from ADP bracis-19054 207 7 fakenewsnet fakenewsnet NOUN bracis-19054 207 8 repository repository NOUN bracis-19054 207 9 ( ( PUNCT bracis-19054 207 10 fnn fnn PROPN bracis-19054 207 11 ) ) PUNCT bracis-19054 207 12     SPACE bracis-19054 208 1 [ [ X bracis-19054 208 2 16 16 NUM bracis-19054 208 3 ] ] X bracis-19054 208 4 , , PUNCT bracis-19054 208 5 which which PRON bracis-19054 208 6 contains contain VERB bracis-19054 208 7 news news NOUN bracis-19054 208 8 of of ADP bracis-19054 208 9 famous famous ADJ bracis-19054 208 10 people people NOUN bracis-19054 208 11 fact fact NOUN bracis-19054 208 12 - - PUNCT bracis-19054 208 13 checked check VERB bracis-19054 208 14 by by ADP bracis-19054 208 15 the the DET bracis-19054 208 16 gossipcop gossipcop PROPN bracis-19054 208 17 website website NOUN bracis-19054 208 18 . . PUNCT bracis-19054 209 1 fakenewsnet fakenewsnet NOUN bracis-19054 209 2 is be AUX bracis-19054 209 3 the the DET bracis-19054 209 4 one one NUM bracis-19054 209 5 with with ADP bracis-19054 209 6 the the DET bracis-19054 209 7 greatest great ADJ bracis-19054 209 8 unbalance unbalance NOUN bracis-19054 209 9 in in ADP bracis-19054 209 10 the the DET bracis-19054 209 11 distribution distribution NOUN bracis-19054 209 12 of of ADP bracis-19054 209 13 classes class NOUN bracis-19054 209 14 . . PUNCT bracis-19054 210 1 the the DET bracis-19054 210 2 last last ADJ bracis-19054 210 3 three three NUM bracis-19054 210 4 datasets dataset NOUN bracis-19054 210 5 are be AUX bracis-19054 210 6 also also ADV bracis-19054 210 7 written write VERB bracis-19054 210 8 in in ADP bracis-19054 210 9 english english PROPN bracis-19054 210 10 . . PUNCT bracis-19054 211 1 the the DET bracis-19054 211 2 news news NOUN bracis-19054 211 3 was be AUX bracis-19054 211 4 taken take VERB bracis-19054 211 5 randomly randomly ADV bracis-19054 211 6 from from ADP bracis-19054 211 7 the the DET bracis-19054 211 8 fakenewscorpus fakenewscorpus NOUN bracis-19054 211 9 datasetfootnote datasetfootnote NOUN bracis-19054 211 10 2 2 NUM bracis-19054 211 11 ( ( PUNCT bracis-19054 211 12 fnc0 fnc0 PROPN bracis-19054 211 13 , , PUNCT bracis-19054 211 14 fnc1 fnc1 PROPN bracis-19054 211 15 , , PUNCT bracis-19054 211 16 fnc2 fnc2 PROPN bracis-19054 211 17 ) ) PUNCT bracis-19054 211 18 , , PUNCT bracis-19054 211 19 an an DET bracis-19054 211 20 open open ADJ bracis-19054 211 21 source source NOUN bracis-19054 211 22 dataset dataset NOUN bracis-19054 211 23 composed compose VERB bracis-19054 211 24 of of ADP bracis-19054 211 25 millions million NOUN bracis-19054 211 26 of of ADP bracis-19054 211 27 news news NOUN bracis-19054 211 28 , , PUNCT bracis-19054 211 29 collected collect VERB bracis-19054 211 30 from from ADP bracis-19054 211 31 1001 1001 NUM bracis-19054 211 32 domains domain NOUN bracis-19054 211 33 . . PUNCT bracis-19054 212 1 fake fake ADJ bracis-19054 212 2 and and CCONJ bracis-19054 212 3 real real ADJ bracis-19054 212 4 news news NOUN bracis-19054 212 5 were be AUX bracis-19054 212 6 selected select VERB bracis-19054 212 7 , , PUNCT bracis-19054 212 8 and and CCONJ bracis-19054 212 9 stopwords stopword NOUN bracis-19054 212 10 were be AUX bracis-19054 212 11 extracted extract VERB bracis-19054 212 12 that that PRON bracis-19054 212 13 could could AUX bracis-19054 212 14 induce induce VERB bracis-19054 212 15 the the DET bracis-19054 212 16 behavior behavior NOUN bracis-19054 212 17 of of ADP bracis-19054 212 18 classification classification NOUN bracis-19054 212 19 algorithms algorithm NOUN bracis-19054 212 20 , , PUNCT bracis-19054 212 21 such such ADJ bracis-19054 212 22 as as ADP bracis-19054 212 23 links link NOUN bracis-19054 212 24 and and CCONJ bracis-19054 212 25 names name NOUN bracis-19054 212 26 of of ADP bracis-19054 212 27 serious serious ADJ bracis-19054 212 28 publication publication NOUN bracis-19054 212 29 vehicles vehicle NOUN bracis-19054 212 30 . . PUNCT bracis-19054 213 1 5.2 5.2 NUM bracis-19054 213 2 experimental experimental ADJ bracis-19054 213 3 setup setup NOUN bracis-19054 213 4 and and CCONJ bracis-19054 213 5     SPACE bracis-19054 213 6 evaluation evaluation NOUN bracis-19054 213 7 criteria criterion NOUN bracis-19054 213 8 this this DET bracis-19054 213 9 section section NOUN bracis-19054 213 10 presents present VERB bracis-19054 213 11 the the DET bracis-19054 213 12 experiment experiment NOUN bracis-19054 213 13 configuration configuration NOUN bracis-19054 213 14 and and CCONJ bracis-19054 213 15 evaluation evaluation NOUN bracis-19054 213 16 criteria criterion NOUN bracis-19054 213 17 for for ADP bracis-19054 213 18 the the DET bracis-19054 213 19 pu pu PROPN bracis-19054 213 20 - - PUNCT bracis-19054 213 21 lp lp PROPN bracis-19054 213 22 and and CCONJ bracis-19054 213 23 the the DET bracis-19054 213 24 baseline baseline ADJ bracis-19054 213 25 algorithms algorithm NOUN bracis-19054 213 26 . . PUNCT bracis-19054 214 1 after after ADP bracis-19054 214 2 pre pre ADJ bracis-19054 214 3 - - ADJ bracis-19054 214 4 processing processing ADJ bracis-19054 214 5 , , PUNCT bracis-19054 214 6 feature feature NOUN bracis-19054 214 7 vectors vector NOUN bracis-19054 214 8 for for ADP bracis-19054 214 9 news news NOUN bracis-19054 214 10 representation representation NOUN bracis-19054 214 11 were be AUX bracis-19054 214 12 obtained obtain VERB bracis-19054 214 13 considering consider VERB bracis-19054 214 14 two two NUM bracis-19054 214 15 strategies strategy NOUN bracis-19054 214 16 : : PUNCT bracis-19054 214 17 ( ( PUNCT bracis-19054 214 18 i i NOUN bracis-19054 214 19 ) ) PUNCT bracis-19054 214 20 a a DET bracis-19054 214 21 traditional traditional ADJ bracis-19054 214 22 bow bow NOUN bracis-19054 214 23 , , PUNCT bracis-19054 214 24 with with ADP bracis-19054 214 25 tf tf PROPN bracis-19054 214 26 - - PUNCT bracis-19054 214 27 idf idf PROPN bracis-19054 214 28 as as ADP bracis-19054 214 29 the the DET bracis-19054 214 30 term term NOUN bracis-19054 214 31 - - PUNCT bracis-19054 214 32 weighting weight VERB bracis-19054 214 33 scheme scheme NOUN bracis-19054 214 34 ; ; PUNCT bracis-19054 214 35 and and CCONJ bracis-19054 214 36 ( ( PUNCT bracis-19054 214 37 ii ii NOUN bracis-19054 214 38 ) ) PUNCT bracis-19054 214 39 d2v d2v NOUN bracis-19054 214 40 ( ( PUNCT bracis-19054 214 41 paragraph paragraph NOUN bracis-19054 214 42 vectors vector NOUN bracis-19054 214 43 ) ) PUNCT bracis-19054 214 44 . . PUNCT bracis-19054 215 1 in in ADP bracis-19054 215 2 d2v d2v PROPN bracis-19054 215 3 we we PRON bracis-19054 215 4 used use VERB bracis-19054 215 5 the the DET bracis-19054 215 6 union union NOUN bracis-19054 215 7 of of ADP bracis-19054 215 8 the the DET bracis-19054 215 9 models model NOUN bracis-19054 215 10 distributed distribute VERB bracis-19054 215 11 memory memory NOUN bracis-19054 215 12 and and CCONJ bracis-19054 215 13 distributed distribute VERB bracis-19054 215 14 bag bag NOUN bracis-19054 215 15 - - PUNCT bracis-19054 215 16 of of ADP bracis-19054 215 17 - - PUNCT bracis-19054 215 18 words word NOUN bracis-19054 215 19 to to PART bracis-19054 215 20 generate generate VERB bracis-19054 215 21 the the DET bracis-19054 215 22 document document NOUN bracis-19054 215 23 embeddings embedding NOUN bracis-19054 215 24 . . PUNCT bracis-19054 216 1 for for ADP bracis-19054 216 2 training train VERB bracis-19054 216 3 each each PRON bracis-19054 216 4 of of ADP bracis-19054 216 5 these these DET bracis-19054 216 6 models model NOUN bracis-19054 216 7 , , PUNCT bracis-19054 216 8 we we PRON bracis-19054 216 9 consider consider VERB bracis-19054 216 10 the the DET bracis-19054 216 11 average average ADJ bracis-19054 216 12 and and CCONJ bracis-19054 216 13 concatenation concatenation NOUN bracis-19054 216 14 of of ADP bracis-19054 216 15 the the DET bracis-19054 216 16 word word NOUN bracis-19054 216 17 vectors vector NOUN bracis-19054 216 18 to to PART bracis-19054 216 19 create create VERB bracis-19054 216 20 the the DET bracis-19054 216 21 hidden hide VERB bracis-19054 216 22 layer layer NOUN bracis-19054 216 23 ’s ’s PART bracis-19054 216 24 output output NOUN bracis-19054 216 25 . . PUNCT bracis-19054 217 1 also also ADV bracis-19054 217 2 , , PUNCT bracis-19054 217 3 we we PRON bracis-19054 217 4 employed employ VERB bracis-19054 217 5 the the DET bracis-19054 217 6 range range NOUN bracis-19054 217 7 of of ADP bracis-19054 217 8 the the DET bracis-19054 217 9 maximum maximum ADJ bracis-19054 217 10 number number NOUN bracis-19054 217 11 of of ADP bracis-19054 217 12 epochs epoch NOUN bracis-19054 217 13 \(\in \(\in VERB bracis-19054 217 14 \{100 \{100 NOUN bracis-19054 217 15 , , PUNCT bracis-19054 217 16 1000\}\ 1000\}\ NOUN bracis-19054 217 17 ) ) PUNCT bracis-19054 217 18 , , PUNCT bracis-19054 217 19 \(\alpha \(\alpha NOUN bracis-19054 217 20 = = SYM bracis-19054 217 21 0.025\ 0.025\ NUM bracis-19054 217 22 ) ) PUNCT bracis-19054 217 23 and and CCONJ bracis-19054 217 24 \(\alpha \(\alpha NOUN bracis-19054 217 25 _ _ PRON bracis-19054 217 26 { { PUNCT bracis-19054 217 27 min min NOUN bracis-19054 217 28 } } PUNCT bracis-19054 217 29 = = ADJ bracis-19054 217 30 0.0001\ 0.0001\ NOUN bracis-19054 217 31 ) ) PUNCT bracis-19054 217 32 , , PUNCT bracis-19054 217 33 number number NOUN bracis-19054 217 34 of of ADP bracis-19054 217 35 dimensions dimension NOUN bracis-19054 217 36 of of ADP bracis-19054 217 37 each each DET bracis-19054 217 38 model model NOUN bracis-19054 217 39 \(= \(= PROPN bracis-19054 217 40 500\ 500\ PROPN bracis-19054 217 41 ) ) PUNCT bracis-19054 217 42 , , PUNCT bracis-19054 217 43 window window NOUN bracis-19054 217 44 size size NOUN bracis-19054 217 45 \(= \(= NOUN bracis-19054 217 46 8\ 8\ NUM bracis-19054 217 47 ) ) PUNCT bracis-19054 217 48 , , PUNCT bracis-19054 217 49 and and CCONJ bracis-19054 217 50 minimum minimum NOUN bracis-19054 217 51 count count NOUN bracis-19054 217 52 \(= \(= NOUN bracis-19054 217 53 1\ 1\ NUM bracis-19054 217 54 ) ) PUNCT bracis-19054 218 1 [ [ X bracis-19054 218 2 8 8 NUM bracis-19054 218 3 ] ] PUNCT bracis-19054 218 4 . . PUNCT bracis-19054 219 1 for for ADP bracis-19054 219 2 the the DET bracis-19054 219 3 k k PROPN bracis-19054 219 4 - - PUNCT bracis-19054 219 5 nn nn ADJ bracis-19054 219 6 matrix matrix NOUN bracis-19054 219 7 used use VERB bracis-19054 219 8 in in ADP bracis-19054 219 9 pu pu PROPN bracis-19054 219 10 - - PUNCT bracis-19054 219 11 lp lp NOUN bracis-19054 219 12 , , PUNCT bracis-19054 219 13 we we PRON bracis-19054 219 14 used use VERB bracis-19054 219 15 \(k \(k NOUN bracis-19054 219 16 = = PUNCT bracis-19054 220 1 [ [ X bracis-19054 220 2 5,6,7]\ 5,6,7]\ NUM bracis-19054 220 3 ) ) PUNCT bracis-19054 220 4 and and CCONJ bracis-19054 220 5 cosine cosine NOUN bracis-19054 220 6 as as ADP bracis-19054 220 7 similarity similarity NOUN bracis-19054 220 8 measure measure NOUN bracis-19054 220 9 . . PUNCT bracis-19054 221 1 for for ADP bracis-19054 221 2 the the DET bracis-19054 221 3 extraction extraction NOUN bracis-19054 221 4 of of ADP bracis-19054 221 5 reliable reliable ADJ bracis-19054 221 6 interest interest NOUN bracis-19054 221 7 and and CCONJ bracis-19054 221 8 non non ADJ bracis-19054 221 9 - - ADJ bracis-19054 221 10 interest interest ADJ bracis-19054 221 11 sets set NOUN bracis-19054 221 12 , , PUNCT bracis-19054 221 13 we we PRON bracis-19054 221 14 use use VERB bracis-19054 221 15 : : PUNCT bracis-19054 221 16 \(m \(m NOUN bracis-19054 221 17 = = SYM bracis-19054 221 18 2\ 2\ NUM bracis-19054 221 19 ) ) PUNCT bracis-19054 221 20 , , PUNCT bracis-19054 221 21 \(\lambda \(\lambda NOUN bracis-19054 222 1 = = PUNCT bracis-19054 223 1 [ [ X bracis-19054 223 2 0.6 0.6 NUM bracis-19054 223 3 , , PUNCT bracis-19054 223 4 0.8]\ 0.8]\ NUM bracis-19054 223 5 ) ) PUNCT bracis-19054 223 6 , , PUNCT bracis-19054 223 7 and and CCONJ bracis-19054 223 8 \(\alpha \(\alpha NOUN bracis-19054 223 9 = = PUNCT bracis-19054 224 1 [ [ X bracis-19054 224 2 0.005 0.005 NUM bracis-19054 224 3 , , PUNCT bracis-19054 224 4 0.01,0.02]\ 0.01,0.02]\ NOUN bracis-19054 224 5 ) ) PUNCT bracis-19054 224 6 . . PUNCT bracis-19054 225 1 these these DET bracis-19054 225 2 values value NOUN bracis-19054 225 3 were be AUX bracis-19054 225 4 chosen choose VERB bracis-19054 225 5 as as SCONJ bracis-19054 225 6 suggested suggest VERB bracis-19054 225 7 in in ADP bracis-19054 225 8 [ [ X bracis-19054 225 9 9 9 NUM bracis-19054 225 10 ] ] PUNCT bracis-19054 225 11 . . PUNCT bracis-19054 226 1 for for ADP bracis-19054 226 2 the the DET bracis-19054 226 3 selection selection NOUN bracis-19054 226 4 of of ADP bracis-19054 226 5 representative representative NOUN bracis-19054 226 6 unigrams unigram NOUN bracis-19054 226 7 and and CCONJ bracis-19054 226 8 bigrams bigram NOUN bracis-19054 226 9 , , PUNCT bracis-19054 226 10 we we PRON bracis-19054 226 11 use use VERB bracis-19054 226 12 \(\ell \(\ell NOUN bracis-19054 226 13 = = SYM bracis-19054 226 14 0.08\ 0.08\ NOUN bracis-19054 226 15 ) ) PUNCT bracis-19054 226 16 . . PUNCT bracis-19054 227 1 the the DET bracis-19054 227 2 parameter parameter NOUN bracis-19054 227 3 \(\ell \(\ell PROPN bracis-19054 227 4 \ \ NOUN bracis-19054 227 5 ) ) PUNCT bracis-19054 227 6 was be AUX bracis-19054 227 7 chosen choose VERB bracis-19054 227 8 after after ADP bracis-19054 227 9 a a DET bracis-19054 227 10 statistical statistical ADJ bracis-19054 227 11 analysis analysis NOUN bracis-19054 227 12 of of ADP bracis-19054 227 13 the the DET bracis-19054 227 14 sample sample NOUN bracis-19054 227 15 , , PUNCT bracis-19054 227 16 indicating indicate VERB bracis-19054 227 17 that that SCONJ bracis-19054 227 18 about about ADV bracis-19054 227 19 25 25 NUM bracis-19054 227 20 % % NOUN bracis-19054 227 21 of of ADP bracis-19054 227 22 bag bag NOUN bracis-19054 227 23 - - PUNCT bracis-19054 227 24 of of ADP bracis-19054 227 25 - - PUNCT bracis-19054 227 26 words word NOUN bracis-19054 227 27 terms term NOUN bracis-19054 227 28 had have VERB bracis-19054 227 29 tf tf ADJ bracis-19054 227 30 - - PUNCT bracis-19054 227 31 idf idf NOUN bracis-19054 227 32 greater great ADJ bracis-19054 227 33 than than ADP bracis-19054 227 34 0.08 0.08 NUM bracis-19054 227 35 . . PUNCT bracis-19054 228 1 for for ADP bracis-19054 228 2 the the DET bracis-19054 228 3 label label NOUN bracis-19054 228 4 propagation propagation NOUN bracis-19054 228 5 stage stage NOUN bracis-19054 228 6 , , PUNCT bracis-19054 228 7 as as SCONJ bracis-19054 228 8 suggested suggest VERB bracis-19054 228 9 in in ADP bracis-19054 228 10 [ [ X bracis-19054 228 11 14 14 NUM bracis-19054 228 12 ] ] PUNCT bracis-19054 228 13 , , PUNCT bracis-19054 228 14 we we PRON bracis-19054 228 15 used use VERB bracis-19054 228 16 a a DET bracis-19054 228 17 convergence convergence NOUN bracis-19054 228 18 threshold threshold NOUN bracis-19054 228 19 = = NOUN bracis-19054 228 20 0.00005 0.00005 NUM bracis-19054 228 21 and and CCONJ bracis-19054 228 22 a a DET bracis-19054 228 23 maximum maximum ADJ bracis-19054 228 24 number number NOUN bracis-19054 228 25 of of ADP bracis-19054 228 26 iterations iteration NOUN bracis-19054 228 27 = = SYM bracis-19054 228 28 1,000 1,000 NUM bracis-19054 228 29 . . PUNCT bracis-19054 229 1 for for ADP bracis-19054 229 2 the the DET bracis-19054 229 3 gnm gnm PROPN bracis-19054 229 4 \(\alpha \(\alpha X bracis-19054 229 5 = = SYM bracis-19054 229 6 \{0.1 \{0.1 PROPN bracis-19054 229 7 , , PUNCT bracis-19054 229 8 0.5\}\ 0.5\}\ NOUN bracis-19054 229 9 ) ) PUNCT bracis-19054 229 10 and and CCONJ bracis-19054 229 11 \(\lambda \(\lambda NOUN bracis-19054 229 12 = = SYM bracis-19054 229 13 1\ 1\ NUM bracis-19054 229 14 ) ) PUNCT bracis-19054 229 15 was be AUX bracis-19054 229 16 considered consider VERB bracis-19054 229 17 . . PUNCT bracis-19054 230 1 a a DET bracis-19054 230 2 10 10 NUM bracis-19054 230 3 - - ADJ bracis-19054 230 4 fold fold ADJ bracis-19054 230 5 cross cross ADJ bracis-19054 230 6 - - ADJ bracis-19054 230 7 validation validation ADJ bracis-19054 230 8 adapted adapt VERB bracis-19054 230 9 to to ADP bracis-19054 230 10 ocl ocl PROPN bracis-19054 230 11 and and CCONJ bracis-19054 230 12 pul pul NOUN bracis-19054 230 13 problems problem NOUN bracis-19054 230 14 was be AUX bracis-19054 230 15 used use VERB bracis-19054 230 16 as as ADP bracis-19054 230 17 a a DET bracis-19054 230 18 validation validation NOUN bracis-19054 230 19 scheme scheme NOUN bracis-19054 230 20 . . PUNCT bracis-19054 231 1 in in ADP bracis-19054 231 2 this this DET bracis-19054 231 3 case case NOUN bracis-19054 231 4 , , PUNCT bracis-19054 231 5 the the DET bracis-19054 231 6 set set NOUN bracis-19054 231 7 of of ADP bracis-19054 231 8 fake fake ADJ bracis-19054 231 9 news news NOUN bracis-19054 231 10 ( ( PUNCT bracis-19054 231 11 \(\mathcal \(\mathcal ADJ bracis-19054 231 12 { { PUNCT bracis-19054 231 13 d}^{+}\ d}^{+}\ NOUN bracis-19054 231 14 ) ) PUNCT bracis-19054 231 15 ) ) PUNCT bracis-19054 231 16 was be AUX bracis-19054 231 17 randomly randomly ADV bracis-19054 231 18 divided divide VERB bracis-19054 231 19 into into ADP bracis-19054 231 20 10 10 NUM bracis-19054 231 21 folds fold NOUN bracis-19054 231 22 . . PUNCT bracis-19054 232 1 in in ADP bracis-19054 232 2 order order NOUN bracis-19054 232 3 to to PART bracis-19054 232 4 simulate simulate VERB bracis-19054 232 5 a a DET bracis-19054 232 6 semi semi ADJ bracis-19054 232 7 - - ADJ bracis-19054 232 8 supervised supervised ADJ bracis-19054 232 9 learning learning NOUN bracis-19054 232 10 environment environment NOUN bracis-19054 232 11 , , PUNCT bracis-19054 232 12 in in ADP bracis-19054 232 13 which which PRON bracis-19054 232 14 the the DET bracis-19054 232 15 number number NOUN bracis-19054 232 16 of of ADP bracis-19054 232 17 labeled label VERB bracis-19054 232 18 examples example NOUN bracis-19054 232 19 is be AUX bracis-19054 232 20 higher high ADJ bracis-19054 232 21 than than ADP bracis-19054 232 22 the the DET bracis-19054 232 23 unlabeled unlabeled ADJ bracis-19054 232 24 examples example NOUN bracis-19054 232 25 , , PUNCT bracis-19054 232 26 i.e. i.e. X bracis-19054 232 27 , , PUNCT bracis-19054 232 28 \(|\mathcal \(|\mathcal PUNCT bracis-19054 232 29 { { PUNCT bracis-19054 232 30 d}^{+}| d}^{+}| PROPN bracis-19054 232 31 \gg \gg PROPN bracis-19054 232 32 |\mathcal |\mathcal PROPN bracis-19054 232 33 { { PUNCT bracis-19054 232 34 d}^u|\ d}^u|\ PROPN bracis-19054 232 35 ) ) PUNCT bracis-19054 232 36 , , PUNCT bracis-19054 232 37 we we PRON bracis-19054 232 38 carried carry VERB bracis-19054 232 39 out out ADP bracis-19054 232 40 different different ADJ bracis-19054 232 41 experiments experiment NOUN bracis-19054 232 42 considering consider VERB bracis-19054 232 43 as as ADP bracis-19054 232 44 labeled label VERB bracis-19054 232 45 data datum NOUN bracis-19054 232 46 1 1 NUM bracis-19054 232 47 , , PUNCT bracis-19054 232 48 2 2 NUM bracis-19054 232 49 or or CCONJ bracis-19054 232 50 3 3 NUM bracis-19054 232 51 folds fold NOUN bracis-19054 232 52 . . PUNCT bracis-19054 233 1 the the DET bracis-19054 233 2 remaining remain VERB bracis-19054 233 3 folds fold NOUN bracis-19054 233 4 and and CCONJ bracis-19054 233 5 the the DET bracis-19054 233 6 real real ADJ bracis-19054 233 7 news news NOUN bracis-19054 233 8 are be AUX bracis-19054 233 9 : : PUNCT bracis-19054 233 10 ( ( PUNCT bracis-19054 233 11 i i NOUN bracis-19054 233 12 ) ) PUNCT bracis-19054 233 13 considered consider VERB bracis-19054 233 14 as as ADP bracis-19054 233 15 test test NOUN bracis-19054 233 16 documents document NOUN bracis-19054 233 17 for for ADP bracis-19054 233 18 the the DET bracis-19054 233 19 ocl ocl PROPN bracis-19054 233 20 algorithm algorithm NOUN bracis-19054 233 21 ; ; PUNCT bracis-19054 233 22 ( ( PUNCT bracis-19054 233 23 ii ii NOUN bracis-19054 233 24 ) ) PUNCT bracis-19054 233 25 considered consider VERB bracis-19054 233 26 as as ADP bracis-19054 233 27 unlabeled unlabeled ADJ bracis-19054 233 28 and and CCONJ bracis-19054 233 29 test test NOUN bracis-19054 233 30 documents document NOUN bracis-19054 233 31 for for ADP bracis-19054 233 32 the the DET bracis-19054 233 33 pul pul NOUN bracis-19054 233 34 algorithms algorithm NOUN bracis-19054 233 35 . . PUNCT bracis-19054 234 1 we we PRON bracis-19054 234 2 proposed propose VERB bracis-19054 234 3 a a DET bracis-19054 234 4 baseline baseline ADJ bracis-19054 234 5 approach approach NOUN bracis-19054 234 6 using use VERB bracis-19054 234 7 binary binary PROPN bracis-19054 234 8 semi semi ADJ bracis-19054 234 9 - - ADJ bracis-19054 234 10 supervised supervised ADJ bracis-19054 234 11 learning learning NOUN bracis-19054 234 12 to to PART bracis-19054 234 13 assess assess VERB bracis-19054 234 14 the the DET bracis-19054 234 15 labeling labeling NOUN bracis-19054 234 16 procedure procedure NOUN bracis-19054 234 17 of of ADP bracis-19054 234 18 reliable reliable ADJ bracis-19054 234 19 - - PUNCT bracis-19054 234 20 interest interest NOUN bracis-19054 234 21 and and CCONJ bracis-19054 234 22 reliable reliable ADJ bracis-19054 234 23 - - PUNCT bracis-19054 234 24 non non ADJ bracis-19054 234 25 - - ADJ bracis-19054 234 26 interest interest ADJ bracis-19054 234 27 examples example NOUN bracis-19054 234 28 in in ADP bracis-19054 234 29 pu pu PROPN bracis-19054 234 30 - - PUNCT bracis-19054 234 31 lp lp NOUN bracis-19054 234 32 . . PUNCT bracis-19054 235 1 for for ADP bracis-19054 235 2 this this DET bracis-19054 235 3 analysis analysis NOUN bracis-19054 235 4 , , PUNCT bracis-19054 235 5 the the DET bracis-19054 235 6 set set NOUN bracis-19054 235 7 of of ADP bracis-19054 235 8 real real ADJ bracis-19054 235 9 news news NOUN bracis-19054 235 10 was be AUX bracis-19054 235 11 randomly randomly ADV bracis-19054 235 12 divided divide VERB bracis-19054 235 13 into into ADP bracis-19054 235 14 ten ten NUM bracis-19054 235 15 subsets subset NOUN bracis-19054 235 16 . . PUNCT bracis-19054 236 1 in in ADP bracis-19054 236 2 the the DET bracis-19054 236 3 cross cross ADJ bracis-19054 236 4 - - ADJ bracis-19054 236 5 validation validation ADJ bracis-19054 236 6 scheme scheme NOUN bracis-19054 236 7 , , PUNCT bracis-19054 236 8 for for SCONJ bracis-19054 236 9 each each DET bracis-19054 236 10 fake fake ADJ bracis-19054 236 11 news news NOUN bracis-19054 236 12 fold fold NOUN bracis-19054 236 13 used use VERB bracis-19054 236 14 to to PART bracis-19054 236 15 train train VERB bracis-19054 236 16 the the DET bracis-19054 236 17 algorithm algorithm NOUN bracis-19054 236 18 , , PUNCT bracis-19054 236 19 one one NUM bracis-19054 236 20 fold fold NOUN bracis-19054 236 21 of of ADP bracis-19054 236 22 real real ADJ bracis-19054 236 23 news news NOUN bracis-19054 236 24 was be AUX bracis-19054 236 25 used use VERB bracis-19054 236 26 . . PUNCT bracis-19054 237 1 from from ADP bracis-19054 237 2 the the DET bracis-19054 237 3 network network NOUN bracis-19054 237 4 obtained obtain VERB bracis-19054 237 5 by by ADP bracis-19054 237 6 \(\mathcal \(\mathcal ADJ bracis-19054 237 7 { { PUNCT bracis-19054 237 8 n}\ n}\ NOUN bracis-19054 237 9 ) ) PUNCT bracis-19054 237 10 in in ADP bracis-19054 237 11 pu pu PROPN bracis-19054 237 12 - - PUNCT bracis-19054 237 13 lp lp NOUN bracis-19054 237 14 , , PUNCT bracis-19054 237 15 and and CCONJ bracis-19054 237 16 considering consider VERB bracis-19054 237 17 the the DET bracis-19054 237 18 training training NOUN bracis-19054 237 19 set set VERB bracis-19054 237 20 as as ADP bracis-19054 237 21 the the DET bracis-19054 237 22 set set NOUN bracis-19054 237 23 of of ADP bracis-19054 237 24 labeled label VERB bracis-19054 237 25 nodes node NOUN bracis-19054 237 26 , , PUNCT bracis-19054 237 27 label label NOUN bracis-19054 237 28 propagation propagation NOUN bracis-19054 237 29 algorithms algorithm NOUN bracis-19054 237 30 infer infer VERB bracis-19054 237 31 the the DET bracis-19054 237 32 class class NOUN bracis-19054 237 33 of of ADP bracis-19054 237 34 the the DET bracis-19054 237 35 remaining remain VERB bracis-19054 237 36 news news NOUN bracis-19054 237 37 from from ADP bracis-19054 237 38 the the DET bracis-19054 237 39 network network NOUN bracis-19054 237 40 . . PUNCT bracis-19054 238 1 we we PRON bracis-19054 238 2 considered consider VERB bracis-19054 238 3 the the DET bracis-19054 238 4 values value NOUN bracis-19054 238 5 of of ADP bracis-19054 238 6 k k NOUN bracis-19054 238 7 ranging range VERB bracis-19054 238 8 in in ADP bracis-19054 238 9 the the DET bracis-19054 238 10 interval interval NOUN bracis-19054 238 11 [ [ X bracis-19054 238 12 5 5 NUM bracis-19054 238 13 , , PUNCT bracis-19054 238 14     SPACE bracis-19054 238 15 7 7 NUM bracis-19054 238 16 ] ] PUNCT bracis-19054 238 17 . . PUNCT bracis-19054 239 1 the the DET bracis-19054 239 2 propagation propagation NOUN bracis-19054 239 3 algorithms algorithm NOUN bracis-19054 239 4 and and CCONJ bracis-19054 239 5 their their PRON bracis-19054 239 6 respective respective ADJ bracis-19054 239 7 configurations configuration NOUN bracis-19054 239 8 are be AUX bracis-19054 239 9 the the DET bracis-19054 239 10 same same ADJ bracis-19054 239 11 as as SCONJ bracis-19054 239 12 the the DET bracis-19054 239 13 experiments experiment NOUN bracis-19054 239 14 performed perform VERB bracis-19054 239 15 with with ADP bracis-19054 239 16 pu pu PROPN bracis-19054 239 17 - - PUNCT bracis-19054 239 18 lp lp NOUN bracis-19054 239 19 . . PUNCT bracis-19054 240 1 as as ADP bracis-19054 240 2 evaluation evaluation NOUN bracis-19054 240 3 measure measure NOUN bracis-19054 240 4 , , PUNCT bracis-19054 240 5 we we PRON bracis-19054 240 6 used use VERB bracis-19054 240 7 \(f_{1 \(f_{1 NOUN bracis-19054 240 8 } } PUNCT bracis-19054 240 9 = = SYM bracis-19054 240 10 ( ( PUNCT bracis-19054 240 11 2 2 NUM bracis-19054 240 12 \cdot \cdot NOUN bracis-19054 240 13 \textit{precision \textit{precision NOUN bracis-19054 240 14 } } PUNCT bracis-19054 240 15 \cdot \cdot NOUN bracis-19054 240 16 \textit{recall})/(\textit{precision}+\textit{recall})\ \textit{recall})/(\textit{precision}+\textit{recall})\ PROPN bracis-19054 240 17 ) ) PUNCT bracis-19054 240 18 considering consider VERB bracis-19054 240 19 the the DET bracis-19054 240 20 fake fake ADJ bracis-19054 240 21 news news NOUN bracis-19054 240 22 as as ADP bracis-19054 240 23 the the DET bracis-19054 240 24 positive positive ADJ bracis-19054 240 25 class class NOUN bracis-19054 240 26 ( ( PUNCT bracis-19054 240 27 interest-\(f_1\))footnote interest-\(f_1\))footnote NOUN bracis-19054 240 28 3 3 NUM bracis-19054 240 29 . . PUNCT bracis-19054 241 1 in in ADP bracis-19054 241 2 the the DET bracis-19054 241 3 next next ADJ bracis-19054 241 4 section section NOUN bracis-19054 241 5 , , PUNCT bracis-19054 241 6 the the DET bracis-19054 241 7 results result NOUN bracis-19054 241 8 of of ADP bracis-19054 241 9 the the DET bracis-19054 241 10 experiments experiment NOUN bracis-19054 241 11 are be AUX bracis-19054 241 12 presented present VERB bracis-19054 241 13 . . PUNCT bracis-19054 242 1 5.3 5.3 NUM bracis-19054 242 2 results result NOUN bracis-19054 242 3 and and CCONJ bracis-19054 242 4     SPACE bracis-19054 242 5 discussions discussion NOUN bracis-19054 242 6 tables table NOUN bracis-19054 242 7     SPACE bracis-19054 242 8 4 4 NUM bracis-19054 242 9 and and CCONJ bracis-19054 242 10 5 5 NUM bracis-19054 242 11 present present VERB bracis-19054 242 12 the the DET bracis-19054 242 13 the the DET bracis-19054 242 14 best good ADJ bracis-19054 242 15 results result NOUN bracis-19054 242 16 for for ADP bracis-19054 242 17 the the DET bracis-19054 242 18 interest-\(f_{1}\ interest-\(f_{1}\ NOUN bracis-19054 242 19 ) ) PUNCT bracis-19054 242 20 that that SCONJ bracis-19054 242 21 each each DET bracis-19054 242 22 network network NOUN bracis-19054 242 23 reached reach VERB bracis-19054 242 24 considering consider VERB bracis-19054 242 25 bow bow VERB bracis-19054 242 26 and and CCONJ bracis-19054 242 27 d2v d2v VERB bracis-19054 242 28 representation representation NOUN bracis-19054 242 29 models model NOUN bracis-19054 242 30 for for ADP bracis-19054 242 31 both both DET bracis-19054 242 32 pu pu PROPN bracis-19054 242 33 - - PUNCT bracis-19054 242 34 lp lp ADJ bracis-19054 242 35 and and CCONJ bracis-19054 242 36 binary binary ADJ bracis-19054 242 37 semi semi ADJ bracis-19054 242 38 - - ADJ bracis-19054 242 39 supervised supervised ADJ bracis-19054 242 40 baseline baseline NOUN bracis-19054 242 41 , , PUNCT bracis-19054 242 42 the the DET bracis-19054 242 43 experimental experimental ADJ bracis-19054 242 44 configuration configuration NOUN bracis-19054 242 45 defined define VERB bracis-19054 242 46 in in ADP bracis-19054 242 47 sect sect NOUN bracis-19054 242 48 . . PUNCT bracis-19054 242 49     SPACE bracis-19054 242 50 5.2 5.2 NUM bracis-19054 242 51 , , PUNCT bracis-19054 242 52 and and CCONJ bracis-19054 242 53 the the DET bracis-19054 242 54 news news NOUN bracis-19054 242 55 datasets dataset NOUN bracis-19054 242 56 . . PUNCT bracis-19054 243 1 due due ADP bracis-19054 243 2 to to ADP bracis-19054 243 3 space space NOUN bracis-19054 243 4 limitations limitation NOUN bracis-19054 243 5 , , PUNCT bracis-19054 243 6 we we PRON bracis-19054 243 7 present present VERB bracis-19054 243 8 the the DET bracis-19054 243 9 results result NOUN bracis-19054 243 10 obtained obtain VERB bracis-19054 243 11 only only ADV bracis-19054 243 12 with with ADP bracis-19054 243 13 the the DET bracis-19054 243 14 gnm gnm PROPN bracis-19054 243 15 algorithm algorithm PROPN bracis-19054 243 16 , , PUNCT bracis-19054 243 17 which which PRON bracis-19054 243 18 obtained obtain VERB bracis-19054 243 19 better well ADJ bracis-19054 243 20 overall overall ADJ bracis-19054 243 21 performance performance NOUN bracis-19054 243 22 . . PUNCT bracis-19054 244 1 the the DET bracis-19054 244 2 complete complete ADJ bracis-19054 244 3 table table NOUN bracis-19054 244 4 of of ADP bracis-19054 244 5 results result NOUN bracis-19054 244 6 is be AUX bracis-19054 244 7 available available ADJ bracis-19054 244 8 in in ADP bracis-19054 244 9 our our PRON bracis-19054 244 10 public public ADJ bracis-19054 244 11 repository repository NOUN bracis-19054 244 12 . . PUNCT bracis-19054 245 1 10 10 NUM bracis-19054 245 2 % % NOUN bracis-19054 245 3 , , PUNCT bracis-19054 245 4 20 20 NUM bracis-19054 245 5 % % NOUN bracis-19054 245 6 , , PUNCT bracis-19054 245 7 and and CCONJ bracis-19054 245 8 30 30 NUM bracis-19054 245 9 % % NOUN bracis-19054 245 10 indicate indicate VERB bracis-19054 245 11 the the DET bracis-19054 245 12 percentage percentage NOUN bracis-19054 245 13 of of ADP bracis-19054 245 14 news news NOUN bracis-19054 245 15 used use VERB bracis-19054 245 16 to to PART bracis-19054 245 17 train train VERB bracis-19054 245 18 the the DET bracis-19054 245 19 algorithms algorithm NOUN bracis-19054 245 20 . . PUNCT bracis-19054 246 1 networks network NOUN bracis-19054 246 2 1 1 NUM bracis-19054 246 3 to to PART bracis-19054 246 4 12 12 NUM bracis-19054 246 5 have have VERB bracis-19054 246 6 a a DET bracis-19054 246 7 combination combination NOUN bracis-19054 246 8 of of ADP bracis-19054 246 9 the the DET bracis-19054 246 10 features feature NOUN bracis-19054 246 11 proposed propose VERB bracis-19054 246 12 in in ADP bracis-19054 246 13 sect sect NOUN bracis-19054 246 14 . . PUNCT bracis-19054 246 15     SPACE bracis-19054 246 16 4.3 4.3 NUM bracis-19054 246 17 ( ( PUNCT bracis-19054 246 18 see see VERB bracis-19054 246 19 table table NOUN bracis-19054 246 20     SPACE bracis-19054 246 21 2 2 NUM bracis-19054 246 22 ) ) PUNCT bracis-19054 246 23 . . PUNCT bracis-19054 247 1 table table NOUN bracis-19054 247 2     SPACE bracis-19054 247 3 6 6 NUM bracis-19054 247 4 presents present VERB bracis-19054 247 5 the the DET bracis-19054 247 6 average average ADJ bracis-19054 247 7 ranking ranking NOUN bracis-19054 247 8 and and CCONJ bracis-19054 247 9 standard standard ADJ bracis-19054 247 10 deviation deviation NOUN bracis-19054 247 11 analysis analysis NOUN bracis-19054 247 12 of of ADP bracis-19054 247 13 the the DET bracis-19054 247 14 proposed propose VERB bracis-19054 247 15 heterogeneous heterogeneous ADJ bracis-19054 247 16 networks network NOUN bracis-19054 247 17 . . PUNCT bracis-19054 248 1 table table NOUN bracis-19054 248 2 4 4 NUM bracis-19054 248 3 . . PUNCT bracis-19054 248 4 interest-\(f_{1}\ interest-\(f_{1}\ ADJ bracis-19054 248 5 ) ) PUNCT bracis-19054 248 6 of of ADP bracis-19054 248 7 the the DET bracis-19054 248 8 pu pu PROPN bracis-19054 248 9 - - PUNCT bracis-19054 248 10 lp lp ADJ bracis-19054 248 11 and and CCONJ bracis-19054 248 12 binary binary ADJ bracis-19054 248 13 baseline baseline PROPN bracis-19054 248 14 approaches approach NOUN bracis-19054 248 15 with with ADP bracis-19054 248 16 bag bag NOUN bracis-19054 248 17 - - PUNCT bracis-19054 248 18 of of ADP bracis-19054 248 19 - - PUNCT bracis-19054 248 20 words word NOUN bracis-19054 248 21 representation representation NOUN bracis-19054 248 22 model model NOUN bracis-19054 248 23 , , PUNCT bracis-19054 248 24 using use VERB bracis-19054 248 25 gnetmine gnetmine NOUN bracis-19054 248 26 as as ADP bracis-19054 248 27 label label NOUN bracis-19054 248 28 propagation.full propagation.full NOUN bracis-19054 248 29 size size NOUN bracis-19054 248 30 table table NOUN bracis-19054 248 31 on on ADP bracis-19054 248 32 network network NOUN bracis-19054 248 33 1 1 NUM bracis-19054 248 34 ( ( PUNCT bracis-19054 248 35 net net NOUN bracis-19054 248 36 1 1 NUM bracis-19054 248 37 ) ) PUNCT bracis-19054 248 38 , , PUNCT bracis-19054 248 39 built build VERB bracis-19054 248 40 only only ADV bracis-19054 248 41 with with ADP bracis-19054 248 42 news news NOUN bracis-19054 248 43 , , PUNCT bracis-19054 248 44 we we PRON bracis-19054 248 45 can can AUX bracis-19054 248 46 see see VERB bracis-19054 248 47 that that SCONJ bracis-19054 248 48 overall overall ADJ bracis-19054 248 49 pu pu PROPN bracis-19054 248 50 - - PUNCT bracis-19054 248 51 lp lp PROPN bracis-19054 248 52 behaves behave VERB bracis-19054 248 53 better well ADV bracis-19054 248 54 than than ADP bracis-19054 248 55 the the DET bracis-19054 248 56 binary binary PROPN bracis-19054 248 57 baseline baseline NOUN bracis-19054 248 58 , , PUNCT bracis-19054 248 59 even even ADV bracis-19054 248 60 using use VERB bracis-19054 248 61 half half NOUN bracis-19054 248 62 of of ADP bracis-19054 248 63 the the DET bracis-19054 248 64 labeled label VERB bracis-19054 248 65 news news NOUN bracis-19054 248 66 . . PUNCT bracis-19054 249 1 when when SCONJ bracis-19054 249 2 exceptions exception NOUN bracis-19054 249 3 occur occur VERB bracis-19054 249 4 , , PUNCT bracis-19054 249 5 the the DET bracis-19054 249 6 results result NOUN bracis-19054 249 7 obtained obtain VERB bracis-19054 249 8 tend tend NOUN bracis-19054 249 9 to to PART bracis-19054 249 10 be be AUX bracis-19054 249 11 very very ADV bracis-19054 249 12 close close ADJ bracis-19054 249 13 , , PUNCT bracis-19054 249 14 with with ADP bracis-19054 249 15 less less ADJ bracis-19054 249 16 than than ADP bracis-19054 249 17 2 2 NUM bracis-19054 249 18 % % NOUN bracis-19054 249 19 of of ADP bracis-19054 249 20 difference difference NOUN bracis-19054 249 21 . . PUNCT bracis-19054 250 1 this this PRON bracis-19054 250 2 demonstrates demonstrate VERB bracis-19054 250 3 that that SCONJ bracis-19054 250 4 identifying identify VERB bracis-19054 250 5 reliable reliable ADJ bracis-19054 250 6 fake fake ADJ bracis-19054 250 7 news news NOUN bracis-19054 250 8 and and CCONJ bracis-19054 250 9 inferring infer VERB bracis-19054 250 10 a a DET bracis-19054 250 11 real real ADJ bracis-19054 250 12 news news NOUN bracis-19054 250 13 set set VERB bracis-19054 250 14 using use VERB bracis-19054 250 15 the the DET bracis-19054 250 16 initially initially ADV bracis-19054 250 17 interest interest NOUN bracis-19054 250 18 labeled label VERB bracis-19054 250 19 set set PROPN bracis-19054 250 20 and and CCONJ bracis-19054 250 21 katz katz PROPN bracis-19054 250 22 index index PROPN bracis-19054 250 23 , , PUNCT bracis-19054 250 24 can can AUX bracis-19054 250 25 be be AUX bracis-19054 250 26 a a DET bracis-19054 250 27 promising promising ADJ bracis-19054 250 28 strategy strategy NOUN bracis-19054 250 29 for for ADP bracis-19054 250 30 news news NOUN bracis-19054 250 31 classification classification NOUN bracis-19054 250 32 . . PUNCT bracis-19054 251 1 in in ADP bracis-19054 251 2 particular particular ADJ bracis-19054 251 3 , , PUNCT bracis-19054 251 4 using use VERB bracis-19054 251 5 only only ADV bracis-19054 251 6 10 10 NUM bracis-19054 251 7 % % NOUN bracis-19054 251 8 of of ADP bracis-19054 251 9 labeled label VERB bracis-19054 251 10 fake fake ADJ bracis-19054 251 11 news news NOUN bracis-19054 251 12 and and CCONJ bracis-19054 251 13 representation representation NOUN bracis-19054 251 14 models model NOUN bracis-19054 251 15 that that PRON bracis-19054 251 16 consider consider VERB bracis-19054 251 17 semantic semantic ADJ bracis-19054 251 18 relations relation NOUN bracis-19054 251 19 such such ADJ bracis-19054 251 20 as as ADP bracis-19054 251 21 d2v d2v PROPN bracis-19054 251 22 , , PUNCT bracis-19054 251 23 can can AUX bracis-19054 251 24 provide provide VERB bracis-19054 251 25 results result NOUN bracis-19054 251 26 as as ADV bracis-19054 251 27 efficient efficient ADJ bracis-19054 251 28 as as ADP bracis-19054 251 29 using use VERB bracis-19054 251 30 larger large ADJ bracis-19054 251 31 training training NOUN bracis-19054 251 32 sets set NOUN bracis-19054 251 33 , , PUNCT bracis-19054 251 34 which which PRON bracis-19054 251 35 favors favor VERB bracis-19054 251 36 the the DET bracis-19054 251 37 dynamic dynamic ADJ bracis-19054 251 38 scenario scenario NOUN bracis-19054 251 39 of of ADP bracis-19054 251 40 news news NOUN bracis-19054 251 41 classification classification NOUN bracis-19054 251 42 . . PUNCT bracis-19054 252 1 including include VERB bracis-19054 252 2 terms term NOUN bracis-19054 252 3 ( ( PUNCT bracis-19054 252 4 net net NOUN bracis-19054 252 5 2 2 NUM bracis-19054 252 6 ) ) PUNCT bracis-19054 252 7 in in ADP bracis-19054 252 8 the the DET bracis-19054 252 9 news news NOUN bracis-19054 252 10 network network NOUN bracis-19054 252 11 tends tend VERB bracis-19054 252 12 to to PART bracis-19054 252 13 improve improve VERB bracis-19054 252 14 classification classification NOUN bracis-19054 252 15 performance performance NOUN bracis-19054 252 16 in in ADP bracis-19054 252 17 general general ADJ bracis-19054 252 18 , , PUNCT bracis-19054 252 19 especially especially ADV bracis-19054 252 20 when when SCONJ bracis-19054 252 21 news news NOUN bracis-19054 252 22 are be AUX bracis-19054 252 23 represented represent VERB bracis-19054 252 24 with with ADP bracis-19054 252 25 bow bow NOUN bracis-19054 252 26 ( ( PUNCT bracis-19054 252 27 table table NOUN bracis-19054 252 28     SPACE bracis-19054 252 29 4 4 NUM bracis-19054 252 30 and and CCONJ bracis-19054 252 31 table table NOUN bracis-19054 252 32     SPACE bracis-19054 252 33 5 5 NUM bracis-19054 252 34 ) ) PUNCT bracis-19054 252 35 . . PUNCT bracis-19054 253 1 table table NOUN bracis-19054 253 2     SPACE bracis-19054 253 3 6 6 NUM bracis-19054 253 4 also also ADV bracis-19054 253 5 shows show VERB bracis-19054 253 6 that that SCONJ bracis-19054 253 7 net net ADJ bracis-19054 253 8 2 2 NUM bracis-19054 253 9 obtained obtain VERB bracis-19054 253 10 a a DET bracis-19054 253 11 better well ADJ bracis-19054 253 12 average average ADJ bracis-19054 253 13 ranking ranking NOUN bracis-19054 253 14 , , PUNCT bracis-19054 253 15 with with ADP bracis-19054 253 16 a a DET bracis-19054 253 17 low low ADJ bracis-19054 253 18 standard standard ADJ bracis-19054 253 19 deviation deviation NOUN bracis-19054 253 20 , , PUNCT bracis-19054 253 21 in in ADP bracis-19054 253 22 relation relation NOUN bracis-19054 253 23 to to ADP bracis-19054 253 24 the the DET bracis-19054 253 25 other other ADJ bracis-19054 253 26 networks network NOUN bracis-19054 253 27 proposed propose VERB bracis-19054 253 28 . . PUNCT bracis-19054 254 1 for for ADP bracis-19054 254 2 bow bow VERB bracis-19054 254 3 with with ADP bracis-19054 254 4 20 20 NUM bracis-19054 254 5 % % NOUN bracis-19054 254 6 and and CCONJ bracis-19054 254 7 30 30 NUM bracis-19054 254 8 % % NOUN bracis-19054 254 9 labeled label VERB bracis-19054 254 10 data datum NOUN bracis-19054 254 11 , , PUNCT bracis-19054 254 12 including include VERB bracis-19054 254 13 also also ADV bracis-19054 254 14 pausality pausality NOUN bracis-19054 254 15 achieved achieve VERB bracis-19054 254 16 better well ADJ bracis-19054 254 17 results result NOUN bracis-19054 254 18 ( ( PUNCT bracis-19054 254 19 net net NOUN bracis-19054 254 20 7 7 NUM bracis-19054 254 21 ) ) PUNCT bracis-19054 254 22 . . PUNCT bracis-19054 255 1 the the DET bracis-19054 255 2 term term NOUN bracis-19054 255 3 selection selection NOUN bracis-19054 255 4 considering consider VERB bracis-19054 255 5 the the DET bracis-19054 255 6 tf tf PROPN bracis-19054 255 7 - - PUNCT bracis-19054 255 8 idf idf PROPN bracis-19054 255 9 measure measure NOUN bracis-19054 255 10 above above ADP bracis-19054 255 11 a a DET bracis-19054 255 12 threshold threshold NOUN bracis-19054 255 13 seems seem VERB bracis-19054 255 14 to to PART bracis-19054 255 15 contribute contribute VERB bracis-19054 255 16 to to ADP bracis-19054 255 17 distinguishing distinguish VERB bracis-19054 255 18 real real ADJ bracis-19054 255 19 and and CCONJ bracis-19054 255 20 fake fake ADJ bracis-19054 255 21 news news NOUN bracis-19054 255 22 by by ADP bracis-19054 255 23 propagation propagation NOUN bracis-19054 255 24 algorithms algorithm NOUN bracis-19054 255 25 , , PUNCT bracis-19054 255 26 increasing increase VERB bracis-19054 255 27 the the DET bracis-19054 255 28 algorithm algorithm NOUN bracis-19054 255 29 ’s ’s PART bracis-19054 255 30 accuracy accuracy NOUN bracis-19054 255 31 . . PUNCT bracis-19054 256 1 table table NOUN bracis-19054 256 2 5 5 NUM bracis-19054 256 3 . . PUNCT bracis-19054 256 4 interest-\(f_{1}\ interest-\(f_{1}\ ADJ bracis-19054 256 5 ) ) PUNCT bracis-19054 256 6 of of ADP bracis-19054 256 7 the the DET bracis-19054 256 8 pu pu PROPN bracis-19054 256 9 - - PUNCT bracis-19054 256 10 lp lp ADJ bracis-19054 256 11 and and CCONJ bracis-19054 256 12 binary binary ADJ bracis-19054 256 13 baseline baseline PROPN bracis-19054 256 14 approaches approach NOUN bracis-19054 256 15 with with ADP bracis-19054 256 16 doc2vec doc2vec NOUN bracis-19054 256 17 representation representation NOUN bracis-19054 256 18 model model NOUN bracis-19054 256 19 , , PUNCT bracis-19054 256 20 using use VERB bracis-19054 256 21 gnetmine gnetmine NOUN bracis-19054 256 22 as as ADP bracis-19054 256 23 label label NOUN bracis-19054 256 24 propagation.full propagation.full NOUN bracis-19054 256 25 size size NOUN bracis-19054 256 26 table table NOUN bracis-19054 256 27 among among ADP bracis-19054 256 28 the the DET bracis-19054 256 29 news news NOUN bracis-19054 256 30 datasets dataset NOUN bracis-19054 257 1 , , PUNCT bracis-19054 257 2 only only ADV bracis-19054 257 3 the the DET bracis-19054 257 4 fnn fnn NOUN bracis-19054 257 5 is be AUX bracis-19054 257 6 unbalanced unbalanced ADJ bracis-19054 257 7 . . PUNCT bracis-19054 258 1 fnn fnn PROPN bracis-19054 258 2 has have VERB bracis-19054 258 3 only only ADV bracis-19054 258 4 celebrity celebrity NOUN bracis-19054 258 5 news news NOUN bracis-19054 258 6 , , PUNCT bracis-19054 258 7 of of ADP bracis-19054 258 8 which which PRON bracis-19054 258 9 fake fake ADJ bracis-19054 258 10 news news NOUN bracis-19054 258 11 is be AUX bracis-19054 258 12 a a DET bracis-19054 258 13 minority minority NOUN bracis-19054 258 14 making make VERB bracis-19054 258 15 up up ADP bracis-19054 258 16 only only ADV bracis-19054 258 17 24.3 24.3 NUM bracis-19054 258 18 % % NOUN bracis-19054 258 19 of of ADP bracis-19054 258 20 the the DET bracis-19054 258 21 dataset dataset NOUN bracis-19054 258 22 . . PUNCT bracis-19054 259 1 for for ADP bracis-19054 259 2 this this DET bracis-19054 259 3 dataset dataset NOUN bracis-19054 259 4 , , PUNCT bracis-19054 259 5 the the DET bracis-19054 259 6 interest-\(f_{1}\ interest-\(f_{1}\ NOUN bracis-19054 259 7 ) ) PUNCT bracis-19054 259 8 always always ADV bracis-19054 259 9 tends tend VERB bracis-19054 259 10 to to PART bracis-19054 259 11 be be AUX bracis-19054 259 12 lower low ADJ bracis-19054 259 13 . . PUNCT bracis-19054 260 1 however however ADV bracis-19054 260 2 , , PUNCT bracis-19054 260 3 we we PRON bracis-19054 260 4 can can AUX bracis-19054 260 5 notice notice VERB bracis-19054 260 6 that that SCONJ bracis-19054 260 7 our our PRON bracis-19054 260 8 proposal proposal NOUN bracis-19054 260 9 performed perform VERB bracis-19054 260 10 better well ADV bracis-19054 260 11 than than ADP bracis-19054 260 12 the the DET bracis-19054 260 13 binary binary PROPN bracis-19054 260 14 baseline baseline NOUN bracis-19054 260 15 , , PUNCT bracis-19054 260 16 which which PRON bracis-19054 260 17 also also ADV bracis-19054 260 18 points point VERB bracis-19054 260 19 out out ADP bracis-19054 260 20 the the DET bracis-19054 260 21 utility utility NOUN bracis-19054 260 22 of of ADP bracis-19054 260 23 pul pul NOUN bracis-19054 260 24 for for ADP bracis-19054 260 25 fake fake ADJ bracis-19054 260 26 news news NOUN bracis-19054 260 27 detection detection NOUN bracis-19054 260 28 . . PUNCT bracis-19054 261 1 lower low ADJ bracis-19054 261 2 results result NOUN bracis-19054 261 3 also also ADV bracis-19054 261 4 occur occur VERB bracis-19054 261 5 in in ADP bracis-19054 261 6 the the DET bracis-19054 261 7 fbr fbr NOUN bracis-19054 261 8 datasets dataset NOUN bracis-19054 261 9 . . PUNCT bracis-19054 262 1 this this DET bracis-19054 262 2 behavior behavior NOUN bracis-19054 262 3 also also ADV bracis-19054 262 4 happened happen VERB bracis-19054 262 5 in in ADP bracis-19054 262 6 [ [ X bracis-19054 262 7 5 5 NUM bracis-19054 262 8 ] ] PUNCT bracis-19054 262 9 , , PUNCT bracis-19054 262 10 in in ADP bracis-19054 262 11 which which PRON bracis-19054 262 12 67 67 NUM bracis-19054 262 13 % % NOUN bracis-19054 262 14 of of ADP bracis-19054 262 15 the the DET bracis-19054 262 16 interest-\(f_1\ interest-\(f_1\ NOUN bracis-19054 262 17 ) ) PUNCT bracis-19054 262 18 was be AUX bracis-19054 262 19 obtained obtain VERB bracis-19054 262 20 using use VERB bracis-19054 262 21 90 90 NUM bracis-19054 262 22 % % NOUN bracis-19054 262 23 labeled label VERB bracis-19054 262 24 fake fake ADJ bracis-19054 262 25 news news NOUN bracis-19054 262 26 and and CCONJ bracis-19054 262 27 one one NUM bracis-19054 262 28 - - PUNCT bracis-19054 262 29 class class NOUN bracis-19054 262 30 learning learning NOUN bracis-19054 262 31 algorithms algorithm NOUN bracis-19054 262 32 . . PUNCT bracis-19054 263 1 the the DET bracis-19054 263 2 hypothesis hypothesis NOUN bracis-19054 263 3 is be AUX bracis-19054 263 4 that that SCONJ bracis-19054 263 5 since since SCONJ bracis-19054 263 6 this this DET bracis-19054 263 7 collection collection NOUN bracis-19054 263 8 in in ADP bracis-19054 263 9 composed compose VERB bracis-19054 263 10 of of ADP bracis-19054 263 11 6 6 NUM bracis-19054 263 12 different different ADJ bracis-19054 263 13 subjects subject NOUN bracis-19054 263 14 , , PUNCT bracis-19054 263 15 fake fake ADJ bracis-19054 263 16 news news NOUN bracis-19054 263 17 from from ADP bracis-19054 263 18 different different ADJ bracis-19054 263 19 subjects subject NOUN bracis-19054 263 20 are be AUX bracis-19054 263 21 spread spread VERB bracis-19054 263 22 along along ADP bracis-19054 263 23 with with ADP bracis-19054 263 24 the the DET bracis-19054 263 25 space space NOUN bracis-19054 263 26 , , PUNCT bracis-19054 263 27 making make VERB bracis-19054 263 28 it it PRON bracis-19054 263 29 difficult difficult ADJ bracis-19054 263 30 to to PART bracis-19054 263 31 infer infer VERB bracis-19054 263 32 reliable reliable ADJ bracis-19054 263 33 negative negative ADJ bracis-19054 263 34 documents document NOUN bracis-19054 263 35 . . PUNCT bracis-19054 264 1 therefore therefore ADV bracis-19054 264 2 , , PUNCT bracis-19054 264 3 we we PRON bracis-19054 264 4 conclude conclude VERB bracis-19054 264 5 that that PRON bracis-19054 264 6 to to PART bracis-19054 264 7 achieve achieve VERB bracis-19054 264 8 more more ADV bracis-19054 264 9 promising promising ADJ bracis-19054 264 10 results result NOUN bracis-19054 264 11 , , PUNCT bracis-19054 264 12 the the DET bracis-19054 264 13 best good ADJ bracis-19054 264 14 way way NOUN bracis-19054 264 15 is be AUX bracis-19054 264 16 to to PART bracis-19054 264 17 classify classify VERB bracis-19054 264 18 the the DET bracis-19054 264 19 news news NOUN bracis-19054 264 20 considering consider VERB bracis-19054 264 21 one one NUM bracis-19054 264 22 subject subject NOUN bracis-19054 264 23 at at ADP bracis-19054 264 24 a a DET bracis-19054 264 25 time time NOUN bracis-19054 264 26 . . PUNCT bracis-19054 265 1 although although SCONJ bracis-19054 265 2 the the DET bracis-19054 265 3 results result NOUN bracis-19054 265 4 achieved achieve VERB bracis-19054 265 5 with with ADP bracis-19054 265 6 these these DET bracis-19054 265 7 datasets dataset NOUN bracis-19054 265 8 are be AUX bracis-19054 265 9 inferior inferior ADJ bracis-19054 265 10 , , PUNCT bracis-19054 265 11 we we PRON bracis-19054 265 12 can can AUX bracis-19054 265 13 observe observe VERB bracis-19054 265 14 that that SCONJ bracis-19054 265 15 the the DET bracis-19054 265 16 addition addition NOUN bracis-19054 265 17 of of ADP bracis-19054 265 18 extra extra ADJ bracis-19054 265 19 information information NOUN bracis-19054 265 20 in in ADP bracis-19054 265 21 the the DET bracis-19054 265 22 heterogeneous heterogeneous ADJ bracis-19054 265 23 network network NOUN bracis-19054 265 24 was be AUX bracis-19054 265 25 beneficial beneficial ADJ bracis-19054 265 26 , , PUNCT bracis-19054 265 27 especially especially ADV bracis-19054 265 28 for for ADP bracis-19054 265 29 the the DET bracis-19054 265 30 baseline baseline NOUN bracis-19054 265 31 , , PUNCT bracis-19054 265 32 which which PRON bracis-19054 265 33 increases increase VERB bracis-19054 265 34 its its PRON bracis-19054 265 35 interest-\(f_{1}\ interest-\(f_{1}\ NOUN bracis-19054 265 36 ) ) PUNCT bracis-19054 265 37 by by ADP bracis-19054 265 38 more more ADJ bracis-19054 265 39 than than ADP bracis-19054 265 40 10 10 NUM bracis-19054 265 41 % % NOUN bracis-19054 265 42 . . PUNCT bracis-19054 266 1 pu pu PROPN bracis-19054 266 2 - - PUNCT bracis-19054 266 3 lp lp PROPN bracis-19054 266 4 obtained obtain VERB bracis-19054 266 5 great great ADJ bracis-19054 266 6 results result NOUN bracis-19054 266 7 for for ADP bracis-19054 266 8 fcn fcn NOUN bracis-19054 266 9 , , PUNCT bracis-19054 266 10 fnc0 fnc0 PROPN bracis-19054 266 11 , , PUNCT bracis-19054 266 12 fnc1 fnc1 PROPN bracis-19054 266 13 and and CCONJ bracis-19054 266 14 fnc2 fnc2 PROPN bracis-19054 266 15 news news PROPN bracis-19054 266 16 datasets dataset NOUN bracis-19054 266 17 . . PUNCT bracis-19054 267 1 using use VERB bracis-19054 267 2 d2v d2v PROPN bracis-19054 267 3 as as ADP bracis-19054 267 4 the the DET bracis-19054 267 5 representation representation NOUN bracis-19054 267 6 model model NOUN bracis-19054 267 7 , , PUNCT bracis-19054 267 8 the the DET bracis-19054 267 9 results result NOUN bracis-19054 267 10 ranged range VERB bracis-19054 267 11 from from ADP bracis-19054 267 12 87 87 NUM bracis-19054 267 13 to to ADP bracis-19054 267 14 92 92 NUM bracis-19054 267 15 % % NOUN bracis-19054 267 16 interest-\(f_{1}\ interest-\(f_{1}\ ADJ bracis-19054 267 17 ) ) PUNCT bracis-19054 267 18 only only ADV bracis-19054 267 19 with with ADP bracis-19054 267 20 news news NOUN bracis-19054 267 21 and and CCONJ bracis-19054 267 22 terms term NOUN bracis-19054 267 23 on on ADP bracis-19054 267 24 the the DET bracis-19054 267 25 network network NOUN bracis-19054 267 26 . . PUNCT bracis-19054 268 1 these these DET bracis-19054 268 2 results result NOUN bracis-19054 268 3 are be AUX bracis-19054 268 4 relevant relevant ADJ bracis-19054 268 5 , , PUNCT bracis-19054 268 6 mainly mainly ADV bracis-19054 268 7 because because SCONJ bracis-19054 268 8 they they PRON bracis-19054 268 9 consider consider VERB bracis-19054 268 10 an an DET bracis-19054 268 11 initially initially ADV bracis-19054 268 12 small small ADJ bracis-19054 268 13 set set NOUN bracis-19054 268 14 of of ADP bracis-19054 268 15 labeled label VERB bracis-19054 268 16 data datum NOUN bracis-19054 268 17 , , PUNCT bracis-19054 268 18 different different ADJ bracis-19054 268 19 from from ADP bracis-19054 268 20 the the DET bracis-19054 268 21 vast vast ADJ bracis-19054 268 22 majority majority NOUN bracis-19054 268 23 of of ADP bracis-19054 268 24 methods method NOUN bracis-19054 268 25 proposed propose VERB bracis-19054 268 26 in in ADP bracis-19054 268 27 the the DET bracis-19054 268 28 literature literature NOUN bracis-19054 268 29 for for ADP bracis-19054 268 30 news news NOUN bracis-19054 268 31 classification classification NOUN bracis-19054 268 32 . . PUNCT bracis-19054 269 1 table table NOUN bracis-19054 269 2 6 6 NUM bracis-19054 269 3 . . PUNCT bracis-19054 269 4 average average ADJ bracis-19054 269 5 ranking ranking NOUN bracis-19054 269 6 and and CCONJ bracis-19054 269 7 standard standard ADJ bracis-19054 269 8 deviation deviation NOUN bracis-19054 269 9 analysis analysis NOUN bracis-19054 269 10 for for ADP bracis-19054 269 11 the the DET bracis-19054 269 12 proposed propose VERB bracis-19054 269 13 heterogeneous heterogeneous ADJ bracis-19054 269 14 networks.full networks.full PRON bracis-19054 269 15 size size NOUN bracis-19054 269 16 table table NOUN bracis-19054 269 17 6 6 NUM bracis-19054 269 18 conclusion conclusion NOUN bracis-19054 269 19 and and CCONJ bracis-19054 269 20     SPACE bracis-19054 269 21 future future ADJ bracis-19054 269 22 work work NOUN bracis-19054 269 23 this this DET bracis-19054 269 24 paper paper NOUN bracis-19054 269 25 proposed propose VERB bracis-19054 269 26 a a DET bracis-19054 269 27 new new ADJ bracis-19054 269 28 approach approach NOUN bracis-19054 269 29 for for ADP bracis-19054 269 30 detecting detect VERB bracis-19054 269 31 fake fake ADJ bracis-19054 269 32 news news NOUN bracis-19054 269 33 based base VERB bracis-19054 269 34 on on ADP bracis-19054 269 35 pu pu PROPN bracis-19054 269 36 - - PUNCT bracis-19054 269 37 lp lp NOUN bracis-19054 269 38 . . PUNCT bracis-19054 270 1 our our PRON bracis-19054 270 2 main main ADJ bracis-19054 270 3 contributions contribution NOUN bracis-19054 270 4 were be AUX bracis-19054 270 5 a a DET bracis-19054 270 6 heterogeneous heterogeneous ADJ bracis-19054 270 7 network network NOUN bracis-19054 270 8 using use VERB bracis-19054 270 9 content content NOUN bracis-19054 270 10 - - PUNCT bracis-19054 270 11 based base VERB bracis-19054 270 12 features feature NOUN bracis-19054 270 13 resulting result VERB bracis-19054 270 14 in in ADP bracis-19054 270 15 a a DET bracis-19054 270 16 domain domain NOUN bracis-19054 270 17 - - PUNCT bracis-19054 270 18 independent independent ADJ bracis-19054 270 19 fake fake ADJ bracis-19054 270 20 news news NOUN bracis-19054 270 21 detection detection NOUN bracis-19054 270 22 approach approach NOUN bracis-19054 270 23 and and CCONJ bracis-19054 270 24 a a DET bracis-19054 270 25 performance performance NOUN bracis-19054 270 26 assessment assessment NOUN bracis-19054 270 27 of of ADP bracis-19054 270 28 different different ADJ bracis-19054 270 29 linguistic linguistic ADJ bracis-19054 270 30 features feature NOUN bracis-19054 270 31 for for ADP bracis-19054 270 32 fake fake ADJ bracis-19054 270 33 news news NOUN bracis-19054 270 34 detection detection NOUN bracis-19054 270 35 . . PUNCT bracis-19054 271 1 pu pu PROPN bracis-19054 271 2 - - PUNCT bracis-19054 271 3 lp lp PROPN bracis-19054 271 4 achieves achieve VERB bracis-19054 271 5 good good ADJ bracis-19054 271 6 classification classification NOUN bracis-19054 271 7 performance performance NOUN bracis-19054 271 8 using use VERB bracis-19054 271 9 a a DET bracis-19054 271 10 low low ADJ bracis-19054 271 11 amount amount NOUN bracis-19054 271 12 of of ADP bracis-19054 271 13 labeled label VERB bracis-19054 271 14 data datum NOUN bracis-19054 271 15 from from ADP bracis-19054 271 16 the the DET bracis-19054 271 17 interest interest NOUN bracis-19054 271 18 class class NOUN bracis-19054 271 19 and and CCONJ bracis-19054 271 20 uses use VERB bracis-19054 271 21 unlabeled unlabele VERB bracis-19054 271 22 data datum NOUN bracis-19054 271 23 to to PART bracis-19054 271 24 improve improve VERB bracis-19054 271 25 classification classification NOUN bracis-19054 271 26 performance performance NOUN bracis-19054 271 27 , , PUNCT bracis-19054 271 28 minimizing minimize VERB bracis-19054 271 29 the the DET bracis-19054 271 30 news news NOUN bracis-19054 271 31 labeling labeling NOUN bracis-19054 271 32 effort effort NOUN bracis-19054 271 33 . . PUNCT bracis-19054 272 1 furthermore furthermore ADV bracis-19054 272 2 , , PUNCT bracis-19054 272 3 since since SCONJ bracis-19054 272 4 pu pu PROPN bracis-19054 272 5 - - PUNCT bracis-19054 272 6 lp lp PROPN bracis-19054 272 7 is be AUX bracis-19054 272 8 a a DET bracis-19054 272 9 pul pul NOUN bracis-19054 272 10 approach approach NOUN bracis-19054 272 11 based base VERB bracis-19054 272 12 on on ADP bracis-19054 272 13 network network NOUN bracis-19054 272 14 , , PUNCT bracis-19054 272 15 it it PRON bracis-19054 272 16 allows allow VERB bracis-19054 272 17 the the DET bracis-19054 272 18 easy easy ADJ bracis-19054 272 19 incorporation incorporation NOUN bracis-19054 272 20 of of ADP bracis-19054 272 21 other other ADJ bracis-19054 272 22 features feature NOUN bracis-19054 272 23 to to PART bracis-19054 272 24 perform perform VERB bracis-19054 272 25 learning learn VERB bracis-19054 272 26 . . PUNCT bracis-19054 273 1 we we PRON bracis-19054 273 2 assessed assess VERB bracis-19054 273 3 the the DET bracis-19054 273 4 performance performance NOUN bracis-19054 273 5 of of ADP bracis-19054 273 6 a a DET bracis-19054 273 7 heterogeneous heterogeneous ADJ bracis-19054 273 8 network network NOUN bracis-19054 273 9 that that PRON bracis-19054 273 10 incorporates incorporate VERB bracis-19054 273 11 combinations combination NOUN bracis-19054 273 12 of of ADP bracis-19054 273 13 four four NUM bracis-19054 273 14 different different ADJ bracis-19054 273 15 features feature NOUN bracis-19054 273 16 based base VERB bracis-19054 273 17 on on ADP bracis-19054 273 18 terms term NOUN bracis-19054 273 19 and and CCONJ bracis-19054 273 20 linguistic linguistic ADJ bracis-19054 273 21 characteristics characteristic NOUN bracis-19054 273 22 . . PUNCT bracis-19054 274 1 the the DET bracis-19054 274 2 results result NOUN bracis-19054 274 3 of of ADP bracis-19054 274 4 our our PRON bracis-19054 274 5 experimental experimental ADJ bracis-19054 274 6 analysis analysis NOUN bracis-19054 274 7 in in ADP bracis-19054 274 8 six six NUM bracis-19054 274 9 datasets dataset NOUN bracis-19054 274 10 from from ADP bracis-19054 274 11 different different ADJ bracis-19054 274 12 languages language NOUN bracis-19054 274 13 and and CCONJ bracis-19054 274 14 subjects subject NOUN bracis-19054 274 15 indicate indicate VERB bracis-19054 274 16 that that SCONJ bracis-19054 274 17 our our PRON bracis-19054 274 18 proposal proposal NOUN bracis-19054 274 19 can can AUX bracis-19054 274 20 provide provide VERB bracis-19054 274 21 good good ADJ bracis-19054 274 22 classification classification NOUN bracis-19054 274 23 performance performance NOUN bracis-19054 274 24 even even ADV bracis-19054 274 25 with with ADP bracis-19054 274 26 a a DET bracis-19054 274 27 small small ADJ bracis-19054 274 28 amount amount NOUN bracis-19054 274 29 labeled label VERB bracis-19054 274 30 fake fake ADJ bracis-19054 274 31 news news NOUN bracis-19054 274 32 . . PUNCT bracis-19054 275 1 incorporating incorporate VERB bracis-19054 275 2 additional additional ADJ bracis-19054 275 3 linguistic linguistic ADJ bracis-19054 275 4 information information NOUN bracis-19054 275 5 such such ADJ bracis-19054 275 6 as as ADP bracis-19054 275 7 representative representative ADJ bracis-19054 275 8 terms term NOUN bracis-19054 275 9 and and CCONJ bracis-19054 275 10 pausality pausality NOUN bracis-19054 275 11 into into ADP bracis-19054 275 12 the the DET bracis-19054 275 13 network network NOUN bracis-19054 275 14 improved improve VERB bracis-19054 275 15 classification classification NOUN bracis-19054 275 16 performance performance NOUN bracis-19054 275 17 . . PUNCT bracis-19054 276 1 our our PRON bracis-19054 276 2 approach approach NOUN bracis-19054 276 3 achieved achieve VERB bracis-19054 276 4 performance performance NOUN bracis-19054 276 5 similar similar ADJ bracis-19054 276 6 to to ADP bracis-19054 276 7 a a DET bracis-19054 276 8 binary binary ADJ bracis-19054 276 9 classifier classifier NOUN bracis-19054 276 10 in in ADP bracis-19054 276 11 all all DET bracis-19054 276 12 datasets dataset NOUN bracis-19054 276 13 , , PUNCT bracis-19054 276 14 even even ADV bracis-19054 276 15 requiring require VERB bracis-19054 276 16 much much ADV bracis-19054 276 17 lesser less ADJ bracis-19054 276 18 effort effort NOUN bracis-19054 276 19 for for ADP bracis-19054 276 20 data datum NOUN bracis-19054 276 21 labeling labeling NOUN bracis-19054 276 22 . . PUNCT bracis-19054 277 1 future future ADJ bracis-19054 277 2 work work NOUN bracis-19054 277 3 intends intend VERB bracis-19054 277 4 to to PART bracis-19054 277 5 consider consider VERB bracis-19054 277 6 more more ADV bracis-19054 277 7 powerful powerful ADJ bracis-19054 277 8 representation representation NOUN bracis-19054 277 9 models model NOUN bracis-19054 277 10 , , PUNCT bracis-19054 277 11 such such ADJ bracis-19054 277 12 as as ADP bracis-19054 277 13 context context NOUN bracis-19054 277 14 - - PUNCT bracis-19054 277 15 based base VERB bracis-19054 277 16 , , PUNCT bracis-19054 277 17 and and CCONJ bracis-19054 277 18 use use VERB bracis-19054 277 19 more more ADV bracis-19054 277 20 efficient efficient ADJ bracis-19054 277 21 label label NOUN bracis-19054 277 22 propagation propagation NOUN bracis-19054 277 23 algorithms algorithm NOUN bracis-19054 277 24 , , PUNCT bracis-19054 277 25 such such ADJ bracis-19054 277 26 as as ADP bracis-19054 277 27 those those PRON bracis-19054 277 28 based base VERB bracis-19054 277 29 on on ADP bracis-19054 277 30 neural neural ADJ bracis-19054 277 31 networks network NOUN bracis-19054 277 32 . . PUNCT bracis-19054 278 1 we we PRON bracis-19054 278 2 also also ADV bracis-19054 278 3 intend intend VERB bracis-19054 278 4 to to PART bracis-19054 278 5 apply apply VERB bracis-19054 278 6 our our PRON bracis-19054 278 7 approach approach NOUN bracis-19054 278 8 to to ADP bracis-19054 278 9 new new ADJ bracis-19054 278 10 datasets dataset NOUN bracis-19054 278 11 , , PUNCT bracis-19054 278 12 which which PRON bracis-19054 278 13 will will AUX bracis-19054 278 14 allow allow VERB bracis-19054 278 15 us we PRON bracis-19054 278 16 to to PART bracis-19054 278 17 perform perform VERB bracis-19054 278 18 statistical statistical ADJ bracis-19054 278 19 significance significance NOUN bracis-19054 278 20 tests test NOUN bracis-19054 278 21 to to PART bracis-19054 278 22 better well ADV bracis-19054 278 23 assess assess VERB bracis-19054 278 24 the the DET bracis-19054 278 25 results result NOUN bracis-19054 278 26 . . PUNCT bracis-19054 279 1 notes note VERB bracis-19054 279 2 1.all 1.all ADJ bracis-19054 279 3 datasets dataset NOUN bracis-19054 279 4 and and CCONJ bracis-19054 279 5 source source NOUN bracis-19054 279 6 codes code NOUN bracis-19054 279 7 used use VERB bracis-19054 279 8 in in ADP bracis-19054 279 9 this this DET bracis-19054 279 10 paper paper NOUN bracis-19054 279 11 are be AUX bracis-19054 279 12 available available ADJ bracis-19054 279 13 in in ADP bracis-19054 279 14 our our PRON bracis-19054 279 15 public public ADJ bracis-19054 279 16 repository repository NOUN bracis-19054 279 17 : : PUNCT bracis-19054 279 18 https://github.com/marianacaravanti/a-heterogeneous-network-based-positive-and-unlabeled-learning-approach-to-detecting-fake-news https://github.com/marianacaravanti/a-heterogeneous-network-based-positive-and-unlabeled-learning-approach-to-detecting-fake-new NOUN bracis-19054 279 19 . . PUNCT bracis-19054 280 1 2.https://github.com/several27/fakenewscorpus 2.https://github.com/several27/fakenewscorpus X bracis-19054 280 2 . . PUNCT bracis-19054 281 1 3.we 3.we NUM bracis-19054 281 2 also also ADV bracis-19054 281 3 evaluate evaluate VERB bracis-19054 281 4 the the DET bracis-19054 281 5 average average NOUN bracis-19054 281 6 of of ADP bracis-19054 281 7 \(f_1\ \(f_1\ NOUN bracis-19054 281 8 ) ) PUNCT bracis-19054 281 9 for for ADP bracis-19054 281 10 both both DET bracis-19054 281 11 classes class NOUN bracis-19054 281 12 ( ( PUNCT bracis-19054 281 13 macro macro NOUN bracis-19054 281 14 - - NOUN bracis-19054 281 15 averaging averaging NOUN bracis-19054 281 16 \(f_{1}\ \(f_{1}\ NOUN bracis-19054 281 17 ) ) PUNCT bracis-19054 281 18 ) ) PUNCT bracis-19054 281 19 . . PUNCT bracis-19054 282 1 due due ADP bracis-19054 282 2 to to ADP bracis-19054 282 3 space space NOUN bracis-19054 282 4 limitations limitation NOUN bracis-19054 282 5 , , PUNCT bracis-19054 282 6 the the DET bracis-19054 282 7 complete complete ADJ bracis-19054 282 8 results result NOUN bracis-19054 282 9 are be AUX bracis-19054 282 10 available available ADJ bracis-19054 282 11 in in ADP bracis-19054 282 12 our our PRON bracis-19054 282 13 public public ADJ bracis-19054 282 14 repository repository NOUN bracis-19054 282 15 : : PUNCT bracis-19054 282 16 https://github.com/marianacaravanti/a-heterogeneous-network-based-positive-and-unlabeled-learning-approach-to-detecting-fake-news/tree/main/results https://github.com/marianacaravanti/a-heterogeneous-network-based-positive-and-unlabeled-learning-approach-to-detecting-fake-news/tree/main/result NOUN bracis-19054 282 17 . . PUNCT bracis-19054 283 1 references reference NOUN bracis-19054 283 2 aggarwal aggarwal PROPN bracis-19054 283 3 , , PUNCT bracis-19054 283 4 c.c c.c PROPN bracis-19054 283 5 . . PROPN bracis-19054 283 6 : : PUNCT bracis-19054 284 1 machine machine NOUN bracis-19054 284 2 learning learn VERB bracis-19054 284 3 for for ADP bracis-19054 284 4 text text NOUN bracis-19054 284 5 . . PUNCT bracis-19054 285 1 springer springer NOUN bracis-19054 285 2 publishing publishing NOUN bracis-19054 285 3 ( ( PUNCT bracis-19054 285 4 2018 2018 NUM bracis-19054 285 5 ) ) PUNCT bracis-19054 285 6 . . PUNCT bracis-19054 286 1 https://doi.org/10.1007/978-3-319-73531-3 https://doi.org/10.1007/978-3-319-73531-3 PRON bracis-19054 286 2 vargas vargas NOUN bracis-19054 286 3 , , PUNCT bracis-19054 286 4 f.a f.a PROPN bracis-19054 286 5 . . PROPN bracis-19054 286 6 , , PUNCT bracis-19054 286 7 pardo pardo PROPN bracis-19054 286 8 , , PUNCT bracis-19054 286 9 t.a.s t.a.s PROPN bracis-19054 286 10 . . PUNCT bracis-19054 286 11 : : PUNCT bracis-19054 286 12 studying study VERB bracis-19054 286 13 dishonest dishonest ADJ bracis-19054 286 14 intentions intention NOUN bracis-19054 286 15 in in ADP bracis-19054 286 16 brazilian brazilian ADJ bracis-19054 286 17 portuguese portuguese ADJ bracis-19054 286 18 texts text NOUN bracis-19054 286 19 . . PUNCT bracis-19054 287 1 arxiv arxiv PROPN bracis-19054 287 2 e e PROPN bracis-19054 287 3 - - NOUN bracis-19054 287 4 prints print NOUN bracis-19054 287 5 ( ( PUNCT bracis-19054 287 6 2020 2020 NUM bracis-19054 287 7 ) ) PUNCT bracis-19054 287 8 google google PROPN bracis-19054 287 9 scholar scholar NOUN bracis-19054 287 10     SPACE bracis-19054 287 11 bekker bekker NOUN bracis-19054 287 12 , , PUNCT bracis-19054 287 13 j. j. PROPN bracis-19054 287 14 , , PUNCT bracis-19054 287 15 davis davis PROPN bracis-19054 287 16 , , PUNCT bracis-19054 287 17 j. j. PROPN bracis-19054 287 18 : : PUNCT bracis-19054 287 19 learning learn VERB bracis-19054 287 20 from from ADP bracis-19054 287 21 positive positive ADJ bracis-19054 287 22 and and CCONJ bracis-19054 287 23 unlabeled unlabeled ADJ bracis-19054 287 24 data datum NOUN bracis-19054 287 25 : : PUNCT bracis-19054 287 26 a a DET bracis-19054 287 27 survey survey NOUN bracis-19054 287 28 . . PUNCT bracis-19054 288 1 mach mach NOUN bracis-19054 288 2 . . PUNCT bracis-19054 289 1 learn learn VERB bracis-19054 289 2 . . PUNCT bracis-19054 290 1 109(4 109(4 NUM bracis-19054 290 2 ) ) PUNCT bracis-19054 290 3 , , PUNCT bracis-19054 290 4 719–760 719–760 NUM bracis-19054 290 5 ( ( PUNCT bracis-19054 290 6 2020 2020 NUM bracis-19054 290 7 ) ) PUNCT bracis-19054 290 8 . . PUNCT bracis-19054 291 1 https://doi.org/10.1007/s10994-020-05877-5 https://doi.org/10.1007/s10994-020-05877-5 NUM bracis-19054 291 2 article article NOUN bracis-19054 291 3     SPACE bracis-19054 291 4 mathscinet mathscinet NOUN bracis-19054 291 5     SPACE bracis-19054 291 6 math math NOUN bracis-19054 291 7     SPACE bracis-19054 291 8 google google PROPN bracis-19054 291 9 scholar scholar NOUN bracis-19054 291 10     SPACE bracis-19054 291 11 bondielli bondielli PROPN bracis-19054 291 12 , , PUNCT bracis-19054 291 13 a. a. NOUN bracis-19054 291 14 , , PUNCT bracis-19054 291 15 marcelloni marcelloni PROPN bracis-19054 291 16 , , PUNCT bracis-19054 291 17 f. f. PROPN bracis-19054 291 18 : : PUNCT bracis-19054 291 19 a a DET bracis-19054 291 20 survey survey NOUN bracis-19054 291 21 on on ADP bracis-19054 291 22 fake fake ADJ bracis-19054 291 23 news news NOUN bracis-19054 291 24 and and CCONJ bracis-19054 291 25 rumour rumour NOUN bracis-19054 291 26 detection detection NOUN bracis-19054 291 27 techniques technique NOUN bracis-19054 291 28 . . PUNCT bracis-19054 292 1 inf inf PROPN bracis-19054 292 2 . . PUNCT bracis-19054 293 1 sci sci PROPN bracis-19054 293 2 . . PROPN bracis-19054 294 1 497 497 NUM bracis-19054 294 2 , , PUNCT bracis-19054 294 3 38–55 38–55 NUM bracis-19054 294 4 ( ( PUNCT bracis-19054 294 5 2019 2019 NUM bracis-19054 294 6 ) ) PUNCT bracis-19054 294 7 article article NOUN bracis-19054 294 8     SPACE bracis-19054 294 9 google google PROPN bracis-19054 294 10 scholar scholar NOUN bracis-19054 294 11     SPACE bracis-19054 294 12 faustini faustini PROPN bracis-19054 294 13 , , PUNCT bracis-19054 294 14 p. p. NOUN bracis-19054 294 15 , , PUNCT bracis-19054 294 16 covões covões PROPN bracis-19054 294 17 , , PUNCT bracis-19054 294 18 t.f t.f PROPN bracis-19054 294 19 . . PUNCT bracis-19054 294 20 : : PUNCT bracis-19054 295 1 fake fake ADJ bracis-19054 295 2 news news NOUN bracis-19054 295 3 detection detection NOUN bracis-19054 295 4 using use VERB bracis-19054 295 5 one one NUM bracis-19054 295 6 - - PUNCT bracis-19054 295 7 class class NOUN bracis-19054 295 8 classification classification NOUN bracis-19054 295 9 . . PUNCT bracis-19054 296 1 in in ADP bracis-19054 296 2 : : PUNCT bracis-19054 296 3 2019 2019 NUM bracis-19054 296 4 8th 8th ADJ bracis-19054 296 5 brazilian brazilian ADJ bracis-19054 296 6 conference conference NOUN bracis-19054 296 7 on on ADP bracis-19054 296 8 intelligent intelligent ADJ bracis-19054 296 9 systems system NOUN bracis-19054 296 10 , , PUNCT bracis-19054 296 11 pp pp ADJ bracis-19054 296 12 . . PUNCT bracis-19054 297 1 592–597 592–597 NUM bracis-19054 297 2 . . PUNCT bracis-19054 298 1 ieee ieee NOUN bracis-19054 298 2 ( ( PUNCT bracis-19054 298 3 2019 2019 NUM bracis-19054 298 4 ) ) PUNCT bracis-19054 298 5 google google NOUN bracis-19054 298 6 scholar scholar NOUN bracis-19054 298 7     SPACE bracis-19054 298 8 greifeneder greifeneder NOUN bracis-19054 298 9 , , PUNCT bracis-19054 298 10 r. r. PROPN bracis-19054 298 11 , , PUNCT bracis-19054 298 12 jaffe jaffe PROPN bracis-19054 298 13 , , PUNCT bracis-19054 298 14 m. m. NOUN bracis-19054 298 15 , , PUNCT bracis-19054 298 16 newman newman PROPN bracis-19054 298 17 , , PUNCT bracis-19054 298 18 e. e. PROPN bracis-19054 298 19 , , PUNCT bracis-19054 298 20 schwarz schwarz PROPN bracis-19054 298 21 , , PUNCT bracis-19054 298 22 n. n. NOUN bracis-19054 298 23 : : PUNCT bracis-19054 298 24 the the DET bracis-19054 298 25 psychology psychology NOUN bracis-19054 298 26 of of ADP bracis-19054 298 27 fake fake ADJ bracis-19054 298 28 news news NOUN bracis-19054 298 29 : : PUNCT bracis-19054 298 30 accepting accept VERB bracis-19054 298 31 , , PUNCT bracis-19054 298 32 sharing sharing NOUN bracis-19054 298 33 , , PUNCT bracis-19054 298 34 and and CCONJ bracis-19054 298 35 correcting correct VERB bracis-19054 298 36 misinformation misinformation NOUN bracis-19054 298 37 . . PUNCT bracis-19054 299 1 routledge routledge PROPN bracis-19054 299 2 , , PUNCT bracis-19054 299 3 milton milton PROPN bracis-19054 299 4 park park PROPN bracis-19054 299 5 ( ( PUNCT bracis-19054 299 6 2021 2021 NUM bracis-19054 299 7 ) ) PUNCT bracis-19054 299 8 google google NOUN bracis-19054 299 9 scholar scholar NOUN bracis-19054 299 10     SPACE bracis-19054 299 11 ji ji PROPN bracis-19054 299 12 , , PUNCT bracis-19054 299 13 m. m. NOUN bracis-19054 299 14 , , PUNCT bracis-19054 299 15 sun sun NOUN bracis-19054 299 16 , , PUNCT bracis-19054 299 17 y. y. NOUN bracis-19054 299 18 , , PUNCT bracis-19054 299 19 danilevsky danilevsky NOUN bracis-19054 299 20 , , PUNCT bracis-19054 299 21 m. m. NOUN bracis-19054 299 22 , , PUNCT bracis-19054 299 23 han han PROPN bracis-19054 299 24 , , PUNCT bracis-19054 299 25 j. j. PROPN bracis-19054 299 26 , , PUNCT bracis-19054 299 27 gao gao PROPN bracis-19054 299 28 , , PUNCT bracis-19054 299 29 j. j. PROPN bracis-19054 299 30 : : PUNCT bracis-19054 299 31 graph graph NOUN bracis-19054 299 32 regularized regularize VERB bracis-19054 299 33 transductive transductive ADJ bracis-19054 299 34 classification classification NOUN bracis-19054 299 35 on on ADP bracis-19054 299 36 heterogeneous heterogeneous ADJ bracis-19054 299 37 information information NOUN bracis-19054 299 38 networks network NOUN bracis-19054 299 39 . . PUNCT bracis-19054 300 1 in in ADP bracis-19054 300 2 : : PUNCT bracis-19054 300 3 european european ADJ bracis-19054 300 4 conference conference NOUN bracis-19054 300 5 on on ADP bracis-19054 300 6 machine machine NOUN bracis-19054 300 7 learning learning NOUN bracis-19054 300 8 and and CCONJ bracis-19054 300 9 knowledge knowledge NOUN bracis-19054 300 10 discovery discovery NOUN bracis-19054 300 11 in in ADP bracis-19054 300 12 databases database NOUN bracis-19054 300 13 , , PUNCT bracis-19054 300 14 pp pp ADV bracis-19054 300 15 . . PUNCT bracis-19054 301 1 570–586 570–586 NUM bracis-19054 301 2 ( ( PUNCT bracis-19054 301 3 2010 2010 NUM bracis-19054 301 4 ) ) PUNCT bracis-19054 301 5 google google PROPN bracis-19054 301 6 scholar scholar NOUN bracis-19054 301 7     SPACE bracis-19054 301 8 le le PROPN bracis-19054 301 9 , , PUNCT bracis-19054 301 10 q. q. PROPN bracis-19054 301 11 , , PUNCT bracis-19054 301 12 mikolov mikolov NOUN bracis-19054 301 13 , , PUNCT bracis-19054 301 14 t. t. PROPN bracis-19054 301 15 : : PUNCT bracis-19054 301 16 distributed distribute VERB bracis-19054 301 17 representations representation NOUN bracis-19054 301 18 of of ADP bracis-19054 301 19 sentences sentence NOUN bracis-19054 301 20 and and CCONJ bracis-19054 301 21 documents document NOUN bracis-19054 301 22 . . PUNCT bracis-19054 302 1 in in ADP bracis-19054 302 2 : : PUNCT bracis-19054 302 3 international international ADJ bracis-19054 302 4 conference conference NOUN bracis-19054 302 5 on on ADP bracis-19054 302 6 machine machine NOUN bracis-19054 302 7 learning learning NOUN bracis-19054 302 8 , , PUNCT bracis-19054 302 9 pp pp ADP bracis-19054 302 10 . . PUNCT bracis-19054 303 1 1188–1196 1188–1196 NUM bracis-19054 303 2 ( ( PUNCT bracis-19054 303 3 2014 2014 NUM bracis-19054 303 4 ) ) PUNCT bracis-19054 303 5 google google PROPN bracis-19054 303 6 scholar scholar NOUN bracis-19054 303 7     SPACE bracis-19054 303 8 ma ma PROPN bracis-19054 303 9 , , PUNCT bracis-19054 303 10 s. s. PROPN bracis-19054 303 11 , , PUNCT bracis-19054 303 12 zhang zhang PROPN bracis-19054 303 13 , , PUNCT bracis-19054 303 14 r. r. PROPN bracis-19054 303 15 : : PUNCT bracis-19054 303 16 pu pu PROPN bracis-19054 303 17 - - PUNCT bracis-19054 303 18 lp lp NOUN bracis-19054 303 19 : : PUNCT bracis-19054 303 20 a a DET bracis-19054 303 21 novel novel ADJ bracis-19054 303 22 approach approach NOUN bracis-19054 303 23 for for ADP bracis-19054 303 24 positive positive ADJ bracis-19054 303 25 and and CCONJ bracis-19054 303 26 unlabeled unlabeled ADJ bracis-19054 303 27 learning learning NOUN bracis-19054 303 28 by by ADP bracis-19054 303 29 label label NOUN bracis-19054 303 30 propagation propagation NOUN bracis-19054 303 31 . . PUNCT bracis-19054 304 1 in in ADP bracis-19054 304 2 : : PUNCT bracis-19054 304 3 2017 2017 NUM bracis-19054 304 4 ieee ieee NOUN bracis-19054 304 5 international international ADJ bracis-19054 304 6 conference conference NOUN bracis-19054 304 7 on on ADP bracis-19054 304 8 multimedia multimedia PROPN bracis-19054 304 9 & & CCONJ bracis-19054 304 10 expo expo NOUN bracis-19054 304 11 workshops workshop NOUN bracis-19054 304 12 ( ( PUNCT bracis-19054 304 13 icmew icmew NOUN bracis-19054 304 14 ) ) PUNCT bracis-19054 304 15 , , PUNCT bracis-19054 304 16 pp pp ADJ bracis-19054 304 17 . . PUNCT bracis-19054 305 1 537–542 537–542 NUM bracis-19054 305 2 . . PUNCT bracis-19054 306 1 ieee ieee NOUN bracis-19054 306 2 ( ( PUNCT bracis-19054 306 3 2017 2017 NUM bracis-19054 306 4 ) ) PUNCT bracis-19054 306 5 google google PROPN bracis-19054 306 6 scholar scholar NOUN bracis-19054 306 7     SPACE bracis-19054 306 8 pennebaker pennebaker NOUN bracis-19054 306 9 , , PUNCT bracis-19054 306 10 j.w j.w PROPN bracis-19054 306 11 . . PROPN bracis-19054 306 12 , , PUNCT bracis-19054 306 13 boyd boyd PROPN bracis-19054 306 14 , , PUNCT bracis-19054 306 15 r.l r.l PROPN bracis-19054 306 16 . . PROPN bracis-19054 306 17 , , PUNCT bracis-19054 306 18 jordan jordan PROPN bracis-19054 306 19 , , PUNCT bracis-19054 306 20 k. k. PROPN bracis-19054 306 21 , , PUNCT bracis-19054 306 22 blackburn blackburn PROPN bracis-19054 306 23 , , PUNCT bracis-19054 306 24 k. k. PROPN bracis-19054 306 25 : : PUNCT bracis-19054 306 26 the the DET bracis-19054 306 27 development development NOUN bracis-19054 306 28 and and CCONJ bracis-19054 306 29 psychometric psychometric ADJ bracis-19054 306 30 properties property NOUN bracis-19054 306 31 of of ADP bracis-19054 306 32 liwc2015 liwc2015 PROPN bracis-19054 306 33 . . PUNCT bracis-19054 307 1 university university PROPN bracis-19054 307 2 of of ADP bracis-19054 307 3 texas texas PROPN bracis-19054 307 4 , , PUNCT bracis-19054 307 5 technical technical ADJ bracis-19054 307 6 report report NOUN bracis-19054 307 7 ( ( PUNCT bracis-19054 307 8 2015 2015 NUM bracis-19054 307 9 ) ) PUNCT bracis-19054 307 10 google google PROPN bracis-19054 307 11 scholar scholar NOUN bracis-19054 307 12     SPACE bracis-19054 307 13 ren ren PROPN bracis-19054 307 14 , , PUNCT bracis-19054 307 15 y. y. PROPN bracis-19054 307 16 , , PUNCT bracis-19054 307 17 wang wang PROPN bracis-19054 307 18 , , PUNCT bracis-19054 307 19 b. b. PROPN bracis-19054 307 20 , , PUNCT bracis-19054 307 21 zhang zhang PROPN bracis-19054 307 22 , , PUNCT bracis-19054 307 23 j. j. PROPN bracis-19054 307 24 , , PUNCT bracis-19054 307 25 chang chang PROPN bracis-19054 307 26 , , PUNCT bracis-19054 307 27 y. y. NOUN bracis-19054 307 28 : : PUNCT bracis-19054 307 29 adversarial adversarial ADJ bracis-19054 307 30 active active ADJ bracis-19054 307 31 learning learning NOUN bracis-19054 307 32 based base VERB bracis-19054 307 33 heterogeneous heterogeneous ADJ bracis-19054 307 34 graph graph NOUN bracis-19054 307 35 neural neural ADJ bracis-19054 307 36 network network NOUN bracis-19054 307 37 for for ADP bracis-19054 307 38 fake fake ADJ bracis-19054 307 39 news news NOUN bracis-19054 307 40 detection detection NOUN bracis-19054 307 41 . . PUNCT bracis-19054 308 1 in in ADP bracis-19054 308 2 : : PUNCT bracis-19054 308 3 2020 2020 NUM bracis-19054 308 4 ieee ieee PROPN bracis-19054 308 5 international international ADJ bracis-19054 308 6 conference conference NOUN bracis-19054 308 7 on on ADP bracis-19054 308 8 data datum NOUN bracis-19054 308 9 mining mining NOUN bracis-19054 308 10 ( ( PUNCT bracis-19054 308 11 icdm icdm NOUN bracis-19054 308 12 ) ) PUNCT bracis-19054 308 13 , , PUNCT bracis-19054 308 14 pp pp ADP bracis-19054 308 15 . . PUNCT bracis-19054 309 1 452–461 452–461 NUM bracis-19054 309 2 . . PUNCT bracis-19054 310 1 ieee ieee NOUN bracis-19054 310 2 ( ( PUNCT bracis-19054 310 3 2020 2020 NUM bracis-19054 310 4 ) ) PUNCT bracis-19054 310 5 google google PROPN bracis-19054 310 6 scholar scholar NOUN bracis-19054 310 7     SPACE bracis-19054 310 8 rossi rossi PROPN bracis-19054 310 9 , , PUNCT bracis-19054 310 10 r.g r.g PROPN bracis-19054 310 11 . . PROPN bracis-19054 310 12 : : PUNCT bracis-19054 310 13 automatic automatic ADJ bracis-19054 310 14 text text NOUN bracis-19054 310 15 classification classification NOUN bracis-19054 310 16 through through ADP bracis-19054 310 17 network network NOUN bracis-19054 310 18 - - PUNCT bracis-19054 310 19 based base VERB bracis-19054 310 20 machine machine NOUN bracis-19054 310 21 learning learning NOUN bracis-19054 310 22 . . PUNCT bracis-19054 311 1 ph.d ph.d PROPN bracis-19054 311 2 . . PUNCT bracis-19054 312 1 thesis thesis NOUN bracis-19054 312 2 , , PUNCT bracis-19054 312 3 university university NOUN bracis-19054 312 4 of of ADP bracis-19054 312 5 são são PROPN bracis-19054 312 6 paulo paulo PROPN bracis-19054 312 7 , , PUNCT bracis-19054 312 8 doctoral doctoral ADJ bracis-19054 312 9 thesis thesis NOUN bracis-19054 312 10 ( ( PUNCT bracis-19054 312 11 2016 2016 NUM bracis-19054 312 12 ) ) PUNCT bracis-19054 312 13 . . PUNCT bracis-19054 313 1 ( ( PUNCT bracis-19054 313 2 in in ADP bracis-19054 313 3 portuguese portuguese ADJ bracis-19054 313 4 ) ) PUNCT bracis-19054 313 5 google google PROPN bracis-19054 313 6 scholar scholar NOUN bracis-19054 313 7     SPACE bracis-19054 313 8 santos santo NOUN bracis-19054 313 9 , , PUNCT bracis-19054 313 10 r.l.s r.l.s NOUN bracis-19054 313 11 . . PROPN bracis-19054 313 12 , , PUNCT bracis-19054 313 13 pardo pardo PROPN bracis-19054 313 14 , , PUNCT bracis-19054 313 15 t.a.s t.a.s PROPN bracis-19054 313 16 . . PUNCT bracis-19054 313 17 : : PUNCT bracis-19054 313 18 fact fact NOUN bracis-19054 313 19 - - PUNCT bracis-19054 313 20 checking checking NOUN bracis-19054 313 21 for for ADP bracis-19054 313 22 portuguese portuguese ADJ bracis-19054 313 23 : : PUNCT bracis-19054 313 24 knowledge knowledge NOUN bracis-19054 313 25 graph graph NOUN bracis-19054 313 26 and and CCONJ bracis-19054 313 27 google google PROPN bracis-19054 313 28 search search NOUN bracis-19054 313 29 - - PUNCT bracis-19054 313 30 based base VERB bracis-19054 313 31 methods method NOUN bracis-19054 313 32 . . PUNCT bracis-19054 314 1 in in ADP bracis-19054 314 2 : : PUNCT bracis-19054 314 3 quaresma quaresma PROPN bracis-19054 314 4 , , PUNCT bracis-19054 314 5 p. p. PROPN bracis-19054 314 6 , , PUNCT bracis-19054 314 7 vieira vieira PROPN bracis-19054 314 8 , , PUNCT bracis-19054 314 9 r. r. PROPN bracis-19054 314 10 , , PUNCT bracis-19054 314 11 aluísio aluísio NOUN bracis-19054 314 12 , , PUNCT bracis-19054 314 13 s. s. PROPN bracis-19054 314 14 , , PUNCT bracis-19054 314 15 moniz moniz PROPN bracis-19054 314 16 , , PUNCT bracis-19054 314 17 h. h. PROPN bracis-19054 314 18 , , PUNCT bracis-19054 314 19 batista batista PROPN bracis-19054 314 20 , , PUNCT bracis-19054 314 21 f. f. PROPN bracis-19054 314 22 , , PUNCT bracis-19054 314 23 gonçalves gonçalves PROPN bracis-19054 314 24 , , PUNCT bracis-19054 314 25 t. t. PROPN bracis-19054 314 26 ( ( PUNCT bracis-19054 314 27 eds eds PROPN bracis-19054 314 28 . . PUNCT bracis-19054 314 29 ) ) PUNCT bracis-19054 314 30 propor propor ADJ bracis-19054 314 31 2020 2020 NUM bracis-19054 314 32 . . PUNCT bracis-19054 315 1 lncs lncs PROPN bracis-19054 315 2 ( ( PUNCT bracis-19054 315 3 lnai lnai ADJ bracis-19054 315 4 ) ) PUNCT bracis-19054 315 5 , , PUNCT bracis-19054 315 6 vol vol NOUN bracis-19054 315 7 . . PROPN bracis-19054 315 8 12037 12037 NUM bracis-19054 315 9 , , PUNCT bracis-19054 315 10 pp pp ADV bracis-19054 315 11 . . PUNCT bracis-19054 316 1 195–205 195–205 NUM bracis-19054 316 2 . . PUNCT bracis-19054 316 3 springer springer NOUN bracis-19054 316 4 , , PUNCT bracis-19054 316 5 cham cham PROPN bracis-19054 316 6 ( ( PUNCT bracis-19054 316 7 2020 2020 NUM bracis-19054 316 8 ) ) PUNCT bracis-19054 316 9 . . PUNCT bracis-19054 317 1 https://doi.org/10.1007/978-3-030-41505-1_19 https://doi.org/10.1007/978-3-030-41505-1_19 PROPN bracis-19054 317 2 chapter chapter NOUN bracis-19054 317 3     SPACE bracis-19054 317 4 google google PROPN bracis-19054 317 5 scholar scholar NOUN bracis-19054 317 6     SPACE bracis-19054 317 7 santos santo NOUN bracis-19054 317 8 , , PUNCT bracis-19054 317 9 b.n b.n PROPN bracis-19054 317 10 . . PROPN bracis-19054 317 11 : : PUNCT bracis-19054 317 12 transductive transductive VERB bracis-19054 317 13 classification classification NOUN bracis-19054 317 14 of of ADP bracis-19054 317 15 events event NOUN bracis-19054 317 16 using use VERB bracis-19054 317 17 heterogeneous heterogeneous ADJ bracis-19054 317 18 networks network NOUN bracis-19054 317 19 . . PUNCT bracis-19054 318 1 master master NOUN bracis-19054 318 2 ’s ’s PART bracis-19054 318 3 thesis thesis PROPN bracis-19054 318 4 federal federal PROPN bracis-19054 318 5 university university PROPN bracis-19054 318 6 of of ADP bracis-19054 318 7 mato mato PROPN bracis-19054 318 8 grosso grosso PROPN bracis-19054 318 9 do do AUX bracis-19054 318 10 sul sul PROPN bracis-19054 318 11 ( ( PUNCT bracis-19054 318 12 2018 2018 NUM bracis-19054 318 13 ) ) PUNCT bracis-19054 318 14 . . PUNCT bracis-19054 319 1 ( ( PUNCT bracis-19054 319 2 in in ADP bracis-19054 319 3 portuguese portuguese ADJ bracis-19054 319 4 ) ) PUNCT bracis-19054 319 5 google google PROPN bracis-19054 319 6 scholar scholar NOUN bracis-19054 319 7     SPACE bracis-19054 319 8 heterogeneous heterogeneous ADJ bracis-19054 319 9 information information NOUN bracis-19054 319 10 network network NOUN bracis-19054 319 11 analysis analysis NOUN bracis-19054 319 12 and and CCONJ bracis-19054 319 13 applications application NOUN bracis-19054 319 14 . . PUNCT bracis-19054 320 1 da da X bracis-19054 320 2 . . PROPN bracis-19054 320 3 springer springer NOUN bracis-19054 320 4 , , PUNCT bracis-19054 320 5 cham cham PROPN bracis-19054 320 6 ( ( PUNCT bracis-19054 320 7 2017 2017 NUM bracis-19054 320 8 ) ) PUNCT bracis-19054 320 9 . . PUNCT bracis-19054 321 1 https://doi.org/10.1007/978-3-319-56212-4_9 https://doi.org/10.1007/978-3-319-56212-4_9 ADJ bracis-19054 321 2 shu shu PROPN bracis-19054 321 3 , , PUNCT bracis-19054 321 4 k. k. PROPN bracis-19054 321 5 , , PUNCT bracis-19054 321 6 mahudeswaran mahudeswaran PROPN bracis-19054 321 7 , , PUNCT bracis-19054 321 8 d. d. PROPN bracis-19054 321 9 , , PUNCT bracis-19054 321 10 wang wang PROPN bracis-19054 321 11 , , PUNCT bracis-19054 321 12 s. s. PROPN bracis-19054 321 13 , , PUNCT bracis-19054 321 14 lee lee PROPN bracis-19054 321 15 , , PUNCT bracis-19054 321 16 d. d. PROPN bracis-19054 321 17 , , PUNCT bracis-19054 321 18 liu liu PROPN bracis-19054 321 19 , , PUNCT bracis-19054 321 20 h. h. PROPN bracis-19054 321 21 : : PUNCT bracis-19054 321 22 fakenewsnet fakenewsnet NOUN bracis-19054 321 23 : : PUNCT bracis-19054 321 24 a a DET bracis-19054 321 25 data data NOUN bracis-19054 321 26 repository repository NOUN bracis-19054 321 27 with with ADP bracis-19054 321 28 news news NOUN bracis-19054 321 29 content content NOUN bracis-19054 321 30 , , PUNCT bracis-19054 321 31 social social ADJ bracis-19054 321 32 context context NOUN bracis-19054 321 33 , , PUNCT bracis-19054 321 34 and and CCONJ bracis-19054 321 35 spatiotemporal spatiotemporal ADJ bracis-19054 321 36 information information NOUN bracis-19054 321 37 for for ADP bracis-19054 321 38 studying study VERB bracis-19054 321 39 fake fake ADJ bracis-19054 321 40 news news NOUN bracis-19054 321 41 on on ADP bracis-19054 321 42 social social ADJ bracis-19054 321 43 media medium NOUN bracis-19054 321 44 . . PUNCT bracis-19054 322 1 big big ADJ bracis-19054 322 2 data datum NOUN bracis-19054 322 3 8(3 8(3 NUM bracis-19054 322 4 ) ) PUNCT bracis-19054 322 5 , , PUNCT bracis-19054 322 6 171–188 171–188 NUM bracis-19054 322 7 ( ( PUNCT bracis-19054 322 8 2020 2020 NUM bracis-19054 322 9 ) ) PUNCT bracis-19054 322 10 article article NOUN bracis-19054 322 11     SPACE bracis-19054 322 12 google google PROPN bracis-19054 322 13 scholar scholar NOUN bracis-19054 322 14     SPACE bracis-19054 322 15 silva silva PROPN bracis-19054 322 16 , , PUNCT bracis-19054 322 17 r.m r.m PROPN bracis-19054 322 18 . . PROPN bracis-19054 322 19 , , PUNCT bracis-19054 322 20 santos santos PROPN bracis-19054 322 21 , , PUNCT bracis-19054 322 22 r.l r.l PROPN bracis-19054 322 23 . . PROPN bracis-19054 322 24 , , PUNCT bracis-19054 322 25 almeida almeida PROPN bracis-19054 322 26 , , PUNCT bracis-19054 322 27 t.a t.a PROPN bracis-19054 322 28 . . PROPN bracis-19054 322 29 , , PUNCT bracis-19054 322 30 pardo pardo PROPN bracis-19054 322 31 , , PUNCT bracis-19054 322 32 t.a t.a PROPN bracis-19054 322 33 . . PROPN bracis-19054 322 34 : : PUNCT bracis-19054 322 35 towards towards ADP bracis-19054 322 36 automatically automatically ADV bracis-19054 322 37 filtering filter VERB bracis-19054 322 38 fake fake ADJ bracis-19054 322 39 news news NOUN bracis-19054 322 40 in in ADP bracis-19054 322 41 portuguese portuguese PROPN bracis-19054 322 42 . . PUNCT bracis-19054 323 1 expert expert PROPN bracis-19054 323 2 syst syst PROPN bracis-19054 323 3 . . PUNCT bracis-19054 324 1 appl appl PROPN bracis-19054 324 2 . . PROPN bracis-19054 325 1 146 146 NUM bracis-19054 325 2 , , PUNCT bracis-19054 325 3 113–199 113–199 NUM bracis-19054 325 4 ( ( PUNCT bracis-19054 325 5 2020 2020 NUM bracis-19054 325 6 ) ) PUNCT bracis-19054 325 7 article article NOUN bracis-19054 325 8     SPACE bracis-19054 325 9 google google PROPN bracis-19054 325 10 scholar scholar NOUN bracis-19054 325 11     SPACE bracis-19054 325 12 singh singh PROPN bracis-19054 325 13 , , PUNCT bracis-19054 325 14 v.k v.k PROPN bracis-19054 325 15 . . PROPN bracis-19054 325 16 , , PUNCT bracis-19054 325 17 ghosh ghosh PROPN bracis-19054 325 18 , , PUNCT bracis-19054 325 19 i. i. PROPN bracis-19054 325 20 , , PUNCT bracis-19054 325 21 sonagara sonagara PROPN bracis-19054 325 22 , , PUNCT bracis-19054 325 23 d. d. PROPN bracis-19054 325 24 : : PUNCT bracis-19054 325 25 detecting detect VERB bracis-19054 325 26 fake fake ADJ bracis-19054 325 27 news news NOUN bracis-19054 325 28 stories story NOUN bracis-19054 325 29 via via ADP bracis-19054 325 30 multimodal multimodal ADJ bracis-19054 325 31 analysis analysis NOUN bracis-19054 325 32 . . PUNCT bracis-19054 326 1 assoc assoc PROPN bracis-19054 326 2 . . PUNCT bracis-19054 326 3 inf inf PROPN bracis-19054 326 4 . . PUNCT bracis-19054 327 1 sci sci PROPN bracis-19054 327 2 . . PROPN bracis-19054 327 3 technol technol PROPN bracis-19054 327 4 . . PUNCT bracis-19054 328 1 72(1 72(1 NOUN bracis-19054 328 2 ) ) PUNCT bracis-19054 328 3 , , PUNCT bracis-19054 328 4 3–17 3–17 PROPN bracis-19054 328 5 ( ( PUNCT bracis-19054 328 6 2021 2021 NUM bracis-19054 328 7 ) ) PUNCT bracis-19054 328 8 article article NOUN bracis-19054 328 9     SPACE bracis-19054 328 10 google google PROPN bracis-19054 328 11 scholar scholar NOUN bracis-19054 328 12     SPACE bracis-19054 328 13 yu yu PROPN bracis-19054 328 14 , , PUNCT bracis-19054 328 15 j. j. PROPN bracis-19054 328 16 , , PUNCT bracis-19054 328 17 huang huang PROPN bracis-19054 328 18 , , PUNCT bracis-19054 328 19 q. q. PROPN bracis-19054 328 20 , , PUNCT bracis-19054 328 21 zhou zhou PROPN bracis-19054 328 22 , , PUNCT bracis-19054 328 23 x. x. PROPN bracis-19054 328 24 , , PUNCT bracis-19054 328 25 sha sha PROPN bracis-19054 328 26 , , PUNCT bracis-19054 328 27 y. y. NOUN bracis-19054 328 28 : : PUNCT bracis-19054 328 29 iarnet iarnet NOUN bracis-19054 328 30 : : PUNCT bracis-19054 328 31 an an DET bracis-19054 328 32 information information NOUN bracis-19054 328 33 aggregating aggregate VERB bracis-19054 328 34 and and CCONJ bracis-19054 328 35 reasoning reasoning NOUN bracis-19054 328 36 network network NOUN bracis-19054 328 37 over over ADP bracis-19054 328 38 heterogeneous heterogeneous ADJ bracis-19054 328 39 graph graph NOUN bracis-19054 328 40 for for ADP bracis-19054 328 41 fake fake ADJ bracis-19054 328 42 news news NOUN bracis-19054 328 43 detection detection NOUN bracis-19054 328 44 . . PUNCT bracis-19054 329 1 in in ADP bracis-19054 329 2 : : PUNCT bracis-19054 329 3 2020 2020 NUM bracis-19054 329 4 international international ADJ bracis-19054 329 5 joint joint ADJ bracis-19054 329 6 conference conference NOUN bracis-19054 329 7 on on ADP bracis-19054 329 8 neural neural ADJ bracis-19054 329 9 networks network NOUN bracis-19054 329 10 ( ( PUNCT bracis-19054 329 11 ijcnn ijcnn PROPN bracis-19054 329 12 ) ) PUNCT bracis-19054 329 13 , , PUNCT bracis-19054 329 14 pp pp ADP bracis-19054 329 15 . . PUNCT bracis-19054 330 1 1–9 1–9 NOUN bracis-19054 330 2 . . NOUN bracis-19054 330 3 ieee ieee PROPN bracis-19054 330 4 ( ( PUNCT bracis-19054 330 5 2020 2020 NUM bracis-19054 330 6 ) ) PUNCT bracis-19054 330 7 google google PROPN bracis-19054 330 8 scholar scholar NOUN bracis-19054 330 9     SPACE bracis-19054 330 10 yu yu PROPN bracis-19054 330 11 , , PUNCT bracis-19054 330 12 s. s. PROPN bracis-19054 330 13 , , PUNCT bracis-19054 330 14 li li PROPN bracis-19054 330 15 , , PUNCT bracis-19054 330 16 c. c. PROPN bracis-19054 330 17 : : PUNCT bracis-19054 330 18 pe pe ADJ bracis-19054 330 19 - - PUNCT bracis-19054 330 20 puc puc NOUN bracis-19054 330 21 : : PUNCT bracis-19054 330 22 a a DET bracis-19054 330 23 graph graph NOUN bracis-19054 330 24 based base VERB bracis-19054 330 25 pu pu PROPN bracis-19054 330 26 - - PUNCT bracis-19054 330 27 learning learn VERB bracis-19054 330 28 approach approach NOUN bracis-19054 330 29 for for ADP bracis-19054 330 30 text text NOUN bracis-19054 330 31 classification classification NOUN bracis-19054 330 32 . . PUNCT bracis-19054 331 1 in in ADP bracis-19054 331 2 : : PUNCT bracis-19054 331 3 perner perner NOUN bracis-19054 331 4 , , PUNCT bracis-19054 331 5 p. p. NOUN bracis-19054 331 6 ( ( PUNCT bracis-19054 331 7 ed ed AUX bracis-19054 331 8 . . PUNCT bracis-19054 331 9 ) ) PUNCT bracis-19054 331 10 mldm mldm VERB bracis-19054 331 11 2007 2007 NUM bracis-19054 331 12 . . PUNCT bracis-19054 332 1 lncs lncs PROPN bracis-19054 332 2 ( ( PUNCT bracis-19054 332 3 lnai lnai ADJ bracis-19054 332 4 ) ) PUNCT bracis-19054 332 5 , , PUNCT bracis-19054 332 6 vol vol NOUN bracis-19054 332 7 . . PROPN bracis-19054 332 8 4571 4571 NUM bracis-19054 332 9 , , PUNCT bracis-19054 332 10 pp pp ADJ bracis-19054 332 11 . . PUNCT bracis-19054 333 1 574–584 574–584 NUM bracis-19054 333 2 . . PUNCT bracis-19054 333 3 springer springer NOUN bracis-19054 333 4 , , PUNCT bracis-19054 333 5 heidelberg heidelberg PROPN bracis-19054 333 6 ( ( PUNCT bracis-19054 333 7 2007 2007 NUM bracis-19054 333 8 ) ) PUNCT bracis-19054 333 9 . . PUNCT bracis-19054 334 1 https://doi.org/10.1007/978-3-540-73499-4_43 https://doi.org/10.1007/978-3-540-73499-4_43 PROPN bracis-19054 334 2 chapter chapter NOUN bracis-19054 334 3     SPACE bracis-19054 334 4 google google PROPN bracis-19054 334 5 scholar scholar NOUN bracis-19054 334 6     SPACE bracis-19054 334 7 zhang zhang PROPN bracis-19054 334 8 , , PUNCT bracis-19054 334 9 j. j. PROPN bracis-19054 334 10 , , PUNCT bracis-19054 334 11 dong dong PROPN bracis-19054 334 12 , , PUNCT bracis-19054 334 13 b. b. PROPN bracis-19054 334 14 , , PUNCT bracis-19054 334 15 philip philip PROPN bracis-19054 334 16 , , PUNCT bracis-19054 334 17 s.y s.y PROPN bracis-19054 334 18 . . PROPN bracis-19054 334 19 : : PUNCT bracis-19054 335 1 deep deep ADJ bracis-19054 335 2 diffusive diffusive ADJ bracis-19054 335 3 neural neural ADJ bracis-19054 335 4 network network NOUN bracis-19054 335 5 based base VERB bracis-19054 335 6 fake fake ADJ bracis-19054 335 7 news news NOUN bracis-19054 335 8 detection detection NOUN bracis-19054 335 9 from from ADP bracis-19054 335 10 heterogeneous heterogeneous ADJ bracis-19054 335 11 social social ADJ bracis-19054 335 12 networks network NOUN bracis-19054 335 13 . . PUNCT bracis-19054 336 1 in in ADP bracis-19054 336 2 : : PUNCT bracis-19054 336 3 big big ADJ bracis-19054 336 4 data datum NOUN bracis-19054 336 5 2019 2019 NUM bracis-19054 336 6 : : PUNCT bracis-19054 336 7 international international ADJ bracis-19054 336 8 conference conference NOUN bracis-19054 336 9 on on ADP bracis-19054 336 10 big big ADJ bracis-19054 336 11 data datum NOUN bracis-19054 336 12 , , PUNCT bracis-19054 336 13 pp pp ADP bracis-19054 336 14 . . PUNCT bracis-19054 337 1 1259–1266 1259–1266 X bracis-19054 337 2 . . PUNCT bracis-19054 338 1 ieee ieee NOUN bracis-19054 338 2 ( ( PUNCT bracis-19054 338 3 2019 2019 NUM bracis-19054 338 4 ) ) PUNCT bracis-19054 338 5 google google PROPN bracis-19054 338 6 scholar scholar NOUN bracis-19054 338 7     SPACE bracis-19054 338 8 zhang zhang PROPN bracis-19054 338 9 , , PUNCT bracis-19054 338 10 x. x. NOUN bracis-19054 338 11 , , PUNCT bracis-19054 338 12 ghorbani ghorbani NOUN bracis-19054 338 13 , , PUNCT bracis-19054 338 14 a.a a.a PROPN bracis-19054 338 15 . . PROPN bracis-19054 338 16 : : PUNCT bracis-19054 338 17 an an DET bracis-19054 338 18 overview overview NOUN bracis-19054 338 19 of of ADP bracis-19054 338 20 online online ADJ bracis-19054 338 21 fake fake ADJ bracis-19054 338 22 news news NOUN bracis-19054 338 23 : : PUNCT bracis-19054 338 24 characterization characterization NOUN bracis-19054 338 25 , , PUNCT bracis-19054 338 26 detection detection NOUN bracis-19054 338 27 , , PUNCT bracis-19054 338 28 and and CCONJ bracis-19054 338 29 discussion discussion NOUN bracis-19054 338 30 . . PUNCT bracis-19054 339 1 inf inf PROPN bracis-19054 339 2 . . PUNCT bracis-19054 339 3 process process NOUN bracis-19054 339 4 . . PUNCT bracis-19054 340 1 manage manage VERB bracis-19054 340 2 . . PUNCT bracis-19054 341 1 57(2 57(2 NUM bracis-19054 341 2 ) ) PUNCT bracis-19054 341 3 , , PUNCT bracis-19054 341 4 102025 102025 NUM bracis-19054 341 5 ( ( PUNCT bracis-19054 341 6 2020 2020 NUM bracis-19054 341 7 ) ) PUNCT bracis-19054 341 8 google google PROPN bracis-19054 341 9 scholar scholar NOUN bracis-19054 341 10     SPACE bracis-19054 341 11 zhu zhu PROPN bracis-19054 341 12 , , PUNCT bracis-19054 341 13 x. x. PROPN bracis-19054 341 14 , , PUNCT bracis-19054 341 15 ghahramani ghahramani PROPN bracis-19054 341 16 , , PUNCT bracis-19054 341 17 z. z. PROPN bracis-19054 341 18 , , PUNCT bracis-19054 341 19 lafferty lafferty PROPN bracis-19054 341 20 , , PUNCT bracis-19054 341 21 j.d j.d PROPN bracis-19054 341 22 . . PROPN bracis-19054 341 23 : : PUNCT bracis-19054 341 24 semi semi ADJ bracis-19054 341 25 - - ADJ bracis-19054 341 26 supervised supervised ADJ bracis-19054 341 27 learning learning NOUN bracis-19054 341 28 using use VERB bracis-19054 341 29 gaussian gaussian ADJ bracis-19054 341 30 fields field NOUN bracis-19054 341 31 and and CCONJ bracis-19054 341 32 harmonic harmonic ADJ bracis-19054 341 33 functions function NOUN bracis-19054 341 34 . . PUNCT bracis-19054 342 1 in in ADP bracis-19054 342 2 : : PUNCT bracis-19054 342 3 proceedings proceeding NOUN bracis-19054 342 4 of of ADP bracis-19054 342 5 the the DET bracis-19054 342 6 20th 20th ADJ bracis-19054 342 7 international international ADJ bracis-19054 342 8 conference conference NOUN bracis-19054 342 9 on on ADP bracis-19054 342 10 machine machine NOUN bracis-19054 342 11 learning learning NOUN bracis-19054 342 12 ( ( PUNCT bracis-19054 342 13 icml-2003 icml-2003 NOUN bracis-19054 342 14 ) ) PUNCT bracis-19054 342 15 , , PUNCT bracis-19054 342 16 pp pp ADP bracis-19054 342 17 . . PUNCT bracis-19054 343 1 912–919 912–919 NUM bracis-19054 343 2 ( ( PUNCT bracis-19054 343 3 2003 2003 NUM bracis-19054 343 4 ) ) PUNCT bracis-19054 343 5 google google PROPN bracis-19054 343 6 scholar scholar NOUN bracis-19054 343 7     SPACE bracis-19054 343 8 download download NOUN bracis-19054 343 9 references reference NOUN bracis-19054 343 10 author author NOUN bracis-19054 343 11 information information NOUN bracis-19054 343 12 authors author NOUN bracis-19054 343 13 and and CCONJ bracis-19054 343 14 affiliations affiliation NOUN bracis-19054 343 15 department department PROPN bracis-19054 343 16 of of ADP bracis-19054 343 17 computer computer NOUN bracis-19054 343 18 science science NOUN bracis-19054 343 19 , , PUNCT bracis-19054 343 20 university university PROPN bracis-19054 343 21 of of ADP bracis-19054 343 22 são são PROPN bracis-19054 343 23 paulo paulo PROPN bracis-19054 343 24 , , PUNCT bracis-19054 343 25 são são PROPN bracis-19054 343 26 carlos carlos PROPN bracis-19054 343 27 , , PUNCT bracis-19054 343 28 brazil brazil PROPN bracis-19054 343 29 mariana mariana PROPN bracis-19054 343 30 c. c. PROPN bracis-19054 343 31 de de PROPN bracis-19054 343 32 souza souza PROPN bracis-19054 343 33 , , PUNCT bracis-19054 343 34     SPACE bracis-19054 343 35 ricardo ricardo PROPN bracis-19054 343 36 m. m. NOUN bracis-19054 343 37 marcacini marcacini PROPN bracis-19054 343 38     SPACE bracis-19054 343 39 & & CCONJ bracis-19054 343 40     SPACE bracis-19054 343 41 solange solange PROPN bracis-19054 343 42 o. o. PROPN bracis-19054 343 43 rezende rezende PROPN bracis-19054 343 44 federal federal PROPN bracis-19054 343 45 university university PROPN bracis-19054 343 46 of of ADP bracis-19054 343 47 mato mato PROPN bracis-19054 343 48 grosso grosso PROPN bracis-19054 343 49 do do AUX bracis-19054 343 50 sul sul PROPN bracis-19054 343 51 , , PUNCT bracis-19054 343 52 campo campo PROPN bracis-19054 343 53 grande grande PROPN bracis-19054 343 54 , , PUNCT bracis-19054 343 55 brazil brazil PROPN bracis-19054 343 56 bruno bruno PROPN bracis-19054 343 57 m. m. PROPN bracis-19054 343 58 nogueira nogueira PROPN bracis-19054 343 59 federal federal PROPN bracis-19054 343 60 university university PROPN bracis-19054 343 61 of of ADP bracis-19054 343 62 mato mato PROPN bracis-19054 343 63 grosso grosso PROPN bracis-19054 343 64 do do AUX bracis-19054 343 65 sul sul PROPN bracis-19054 343 66 , , PUNCT bracis-19054 343 67 três três PROPN bracis-19054 343 68 lagoas lagoas PROPN bracis-19054 343 69 , , PUNCT bracis-19054 343 70 brazil brazil PROPN bracis-19054 343 71 rafael rafael PROPN bracis-19054 343 72 g. g. PROPN bracis-19054 343 73 rossi rossi PROPN bracis-19054 343 74 authors author NOUN bracis-19054 343 75 mariana mariana PROPN bracis-19054 343 76 c. c. PROPN bracis-19054 343 77 de de PROPN bracis-19054 343 78 souzaview souzaview PROPN bracis-19054 343 79 author author NOUN bracis-19054 343 80 publications publication NOUN bracis-19054 343 81 search search NOUN bracis-19054 343 82 author author NOUN bracis-19054 343 83 on on ADP bracis-19054 343 84 : : PUNCT bracis-19054 343 85 pubmed pubmed PROPN bracis-19054 343 86     SPACE bracis-19054 343 87 google google PROPN bracis-19054 343 88 scholar scholar PROPN bracis-19054 343 89 bruno bruno PROPN bracis-19054 343 90 m. m. PROPN bracis-19054 343 91 nogueiraview nogueiraview PROPN bracis-19054 343 92 author author NOUN bracis-19054 343 93 publications publication VERB bracis-19054 343 94 search search NOUN bracis-19054 343 95 author author NOUN bracis-19054 343 96 on on ADP bracis-19054 343 97 : : PUNCT bracis-19054 343 98 pubmed pubmed PROPN bracis-19054 343 99     SPACE bracis-19054 343 100 google google PROPN bracis-19054 343 101 scholar scholar PROPN bracis-19054 343 102 rafael rafael PROPN bracis-19054 343 103 g. g. PROPN bracis-19054 343 104 rossiview rossiview PROPN bracis-19054 343 105 author author NOUN bracis-19054 343 106 publications publication NOUN bracis-19054 343 107 search search NOUN bracis-19054 343 108 author author NOUN bracis-19054 343 109 on on ADP bracis-19054 343 110 : : PUNCT bracis-19054 343 111 pubmed pubmed PROPN bracis-19054 343 112     SPACE bracis-19054 343 113 google google PROPN bracis-19054 343 114 scholar scholar PROPN bracis-19054 343 115 ricardo ricardo PROPN bracis-19054 343 116 m. m. PROPN bracis-19054 343 117 marcaciniview marcaciniview PROPN bracis-19054 343 118 author author NOUN bracis-19054 343 119 publications publication NOUN bracis-19054 343 120 search search NOUN bracis-19054 343 121 author author NOUN bracis-19054 343 122 on on ADP bracis-19054 343 123 : : PUNCT bracis-19054 343 124 pubmed pubmed PROPN bracis-19054 343 125     SPACE bracis-19054 343 126 google google PROPN bracis-19054 343 127 scholar scholar PROPN bracis-19054 343 128 solange solange PROPN bracis-19054 343 129 o. o. PROPN bracis-19054 343 130 rezendeview rezendeview PROPN bracis-19054 343 131 author author NOUN bracis-19054 343 132 publications publication NOUN bracis-19054 343 133 search search NOUN bracis-19054 343 134 author author NOUN bracis-19054 343 135 on on ADP bracis-19054 343 136 : : PUNCT bracis-19054 343 137 pubmed pubmed PROPN bracis-19054 343 138     SPACE bracis-19054 343 139 google google PROPN bracis-19054 343 140 scholar scholar NOUN bracis-19054 343 141 corresponding correspond VERB bracis-19054 343 142 author author NOUN bracis-19054 343 143 correspondence correspondence NOUN bracis-19054 343 144 to to ADP bracis-19054 343 145 mariana mariana PROPN bracis-19054 343 146 c. c. PROPN bracis-19054 343 147 de de PROPN bracis-19054 343 148 souza souza PROPN bracis-19054 343 149 . . PUNCT bracis-19054 344 1 editor editor NOUN bracis-19054 344 2 information information NOUN bracis-19054 344 3 editors editor NOUN bracis-19054 344 4 and and CCONJ bracis-19054 344 5 affiliations affiliation NOUN bracis-19054 344 6 universidade universidade PROPN bracis-19054 344 7 federal federal PROPN bracis-19054 344 8 de de X bracis-19054 344 9 sergipe sergipe PROPN bracis-19054 344 10 , , PUNCT bracis-19054 344 11 são são NOUN bracis-19054 344 12 cristóvão cristóvão PROPN bracis-19054 344 13 , , PUNCT bracis-19054 344 14 brazil brazil PROPN bracis-19054 344 15 andré andré PROPN bracis-19054 344 16 britto britto PROPN bracis-19054 344 17 universidade universidade PROPN bracis-19054 344 18 de de PROPN bracis-19054 344 19 são são PROPN bracis-19054 344 20 paulo paulo PROPN bracis-19054 344 21 , , PUNCT bracis-19054 344 22 são são PROPN bracis-19054 344 23 paulo paulo PROPN bracis-19054 344 24 , , PUNCT bracis-19054 344 25 brazil brazil PROPN bracis-19054 344 26 karina karina PROPN bracis-19054 344 27 valdivia valdivia PROPN bracis-19054 344 28 delgado delgado VERB bracis-19054 344 29 rights right NOUN bracis-19054 344 30 and and CCONJ bracis-19054 344 31 permissions permission NOUN bracis-19054 344 32 reprints reprint NOUN bracis-19054 344 33 and and CCONJ bracis-19054 344 34 permissions permission VERB bracis-19054 344 35 copyright copyright NOUN bracis-19054 344 36 information information NOUN bracis-19054 344 37 © © PROPN bracis-19054 344 38 2021 2021 NUM bracis-19054 344 39 springer springer NOUN bracis-19054 344 40 nature nature NOUN bracis-19054 344 41 switzerland switzerland PROPN bracis-19054 344 42 ag ag PROPN bracis-19054 344 43 about about ADP bracis-19054 344 44 this this DET bracis-19054 344 45 paper paper NOUN bracis-19054 344 46 cite cite VERB bracis-19054 344 47 this this DET bracis-19054 344 48 paper paper NOUN bracis-19054 344 49 de de ADP bracis-19054 344 50 souza souza PROPN bracis-19054 344 51 , , PUNCT bracis-19054 344 52 m.c m.c PROPN bracis-19054 344 53 . . PROPN bracis-19054 344 54 , , PUNCT bracis-19054 344 55 nogueira nogueira PROPN bracis-19054 344 56 , , PUNCT bracis-19054 344 57 b.m b.m PROPN bracis-19054 344 58 . . PROPN bracis-19054 344 59 , , PUNCT bracis-19054 344 60 rossi rossi PROPN bracis-19054 344 61 , , PUNCT bracis-19054 344 62 r.g r.g PROPN bracis-19054 344 63 . . PROPN bracis-19054 344 64 , , PUNCT bracis-19054 344 65 marcacini marcacini PROPN bracis-19054 344 66 , , PUNCT bracis-19054 344 67 r.m r.m PROPN bracis-19054 344 68 . . PROPN bracis-19054 344 69 , , PUNCT bracis-19054 344 70 rezende rezende PROPN bracis-19054 344 71 , , PUNCT bracis-19054 344 72 s.o s.o PROPN bracis-19054 344 73 . . PROPN bracis-19054 344 74 ( ( PUNCT bracis-19054 344 75 2021 2021 NUM bracis-19054 344 76 ) ) PUNCT bracis-19054 344 77 . . PUNCT bracis-19054 345 1 a a DET bracis-19054 345 2 heterogeneous heterogeneous ADJ bracis-19054 345 3 network network NOUN bracis-19054 345 4 - - PUNCT bracis-19054 345 5 based base VERB bracis-19054 345 6 positive positive ADJ bracis-19054 345 7 and and CCONJ bracis-19054 345 8 unlabeled unlabeled ADJ bracis-19054 345 9 learning learn VERB bracis-19054 345 10 approach approach NOUN bracis-19054 345 11 to to PART bracis-19054 345 12 detect detect VERB bracis-19054 345 13 fake fake ADJ bracis-19054 345 14 news news NOUN bracis-19054 345 15 . . PUNCT bracis-19054 346 1 in in ADP bracis-19054 346 2 : : PUNCT bracis-19054 346 3 britto britto PROPN bracis-19054 346 4 , , PUNCT bracis-19054 346 5 a. a. PROPN bracis-19054 346 6 , , PUNCT bracis-19054 346 7 valdivia valdivia PROPN bracis-19054 346 8 delgado delgado PROPN bracis-19054 346 9 , , PUNCT bracis-19054 346 10 k. k. PROPN bracis-19054 346 11 ( ( PUNCT bracis-19054 346 12 eds eds PROPN bracis-19054 346 13 ) ) PUNCT bracis-19054 346 14 intelligent intelligent ADJ bracis-19054 346 15 systems system NOUN bracis-19054 346 16 . . PUNCT bracis-19054 347 1 bracis bracis PROPN bracis-19054 347 2 2021 2021 NUM bracis-19054 347 3 . . PUNCT bracis-19054 348 1 lecture lecture NOUN bracis-19054 348 2 notes note NOUN bracis-19054 348 3 in in ADP bracis-19054 348 4 computer computer NOUN bracis-19054 348 5 science science NOUN bracis-19054 348 6 ( ( PUNCT bracis-19054 348 7 ) ) PUNCT bracis-19054 348 8 , , PUNCT bracis-19054 348 9 vol vol NOUN bracis-19054 348 10 13074 13074 NUM bracis-19054 348 11 . . PUNCT bracis-19054 349 1 springer springer NOUN bracis-19054 349 2 , , PUNCT bracis-19054 349 3 cham cham PROPN bracis-19054 349 4 . . PUNCT bracis-19054 350 1 https://doi.org/10.1007/978-3-030-91699-2_1 https://doi.org/10.1007/978-3-030-91699-2_1 X bracis-19054 350 2 download download PROPN bracis-19054 350 3 citation citation NOUN bracis-19054 350 4 .ris .ris PUNCT bracis-19054 350 5 .enw .enw PROPN bracis-19054 350 6 .bib .bib PUNCT bracis-19054 351 1 doi doi PROPN bracis-19054 351 2 : : PUNCT bracis-19054 351 3 https://doi.org/10.1007/978-3-030-91699-2_1 https://doi.org/10.1007/978-3-030-91699-2_1 X bracis-19054 351 4 published publish VERB bracis-19054 351 5 : : PUNCT bracis-19054 351 6 28 28 NUM bracis-19054 351 7 november november PROPN bracis-19054 351 8 2021 2021 NUM bracis-19054 351 9 publisher publisher NOUN bracis-19054 351 10 name name NOUN bracis-19054 351 11 : : PUNCT bracis-19054 351 12 springer springer NOUN bracis-19054 351 13 , , PUNCT bracis-19054 351 14 cham cham PROPN bracis-19054 351 15 print print PROPN bracis-19054 351 16 isbn isbn PROPN bracis-19054 351 17 : : PUNCT bracis-19054 351 18 978 978 NUM bracis-19054 351 19 - - SYM bracis-19054 351 20 3 3 NUM bracis-19054 351 21 - - PUNCT bracis-19054 351 22 030 030 NUM bracis-19054 351 23 - - PUNCT bracis-19054 351 24 91698 91698 NUM bracis-19054 351 25 - - SYM bracis-19054 351 26 5 5 NUM bracis-19054 351 27 online online ADJ bracis-19054 351 28 isbn isbn NOUN bracis-19054 351 29 : : PUNCT bracis-19054 351 30 978 978 NUM bracis-19054 351 31 - - SYM bracis-19054 351 32 3 3 NUM bracis-19054 351 33 - - PUNCT bracis-19054 351 34 030 030 NUM bracis-19054 351 35 - - PUNCT bracis-19054 351 36 91699 91699 NUM bracis-19054 351 37 - - SYM bracis-19054 351 38 2 2 NUM bracis-19054 351 39 ebook ebook NOUN bracis-19054 351 40 packages package NOUN bracis-19054 351 41 : : PUNCT bracis-19054 351 42 computer computer NOUN bracis-19054 351 43 sciencecomputer sciencecomputer NOUN bracis-19054 351 44 science science NOUN bracis-19054 351 45 ( ( PUNCT bracis-19054 351 46 r0 r0 NOUN bracis-19054 351 47 ) ) PUNCT bracis-19054 351 48 share share VERB bracis-19054 351 49 this this DET bracis-19054 351 50 paper paper NOUN bracis-19054 351 51 anyone anyone PRON bracis-19054 351 52 you you PRON bracis-19054 351 53 share share VERB bracis-19054 351 54 the the DET bracis-19054 351 55 following follow VERB bracis-19054 351 56 link link NOUN bracis-19054 351 57 with with ADP bracis-19054 351 58 will will AUX bracis-19054 351 59 be be AUX bracis-19054 351 60 able able ADJ bracis-19054 351 61 to to PART bracis-19054 351 62 read read VERB bracis-19054 351 63 this this DET bracis-19054 351 64 content content NOUN bracis-19054 351 65 : : PUNCT bracis-19054 351 66 get get VERB bracis-19054 351 67 shareable shareable ADJ bracis-19054 351 68 linksorry linksorry NOUN bracis-19054 351 69 , , PUNCT bracis-19054 351 70 a a DET bracis-19054 351 71 shareable shareable ADJ bracis-19054 351 72 link link NOUN bracis-19054 351 73 is be AUX bracis-19054 351 74 not not PART bracis-19054 351 75 currently currently ADV bracis-19054 351 76 available available ADJ bracis-19054 351 77 for for ADP bracis-19054 351 78 this this DET bracis-19054 351 79 article article NOUN bracis-19054 351 80 . . PUNCT bracis-19054 352 1 copy copy VERB bracis-19054 352 2 shareable shareable ADJ bracis-19054 352 3 link link NOUN bracis-19054 352 4 to to PART bracis-19054 352 5 clipboard clipboard NOUN bracis-19054 352 6 provided provide VERB bracis-19054 352 7 by by ADP bracis-19054 352 8 the the DET bracis-19054 352 9 springer springer NOUN bracis-19054 352 10 nature nature PROPN bracis-19054 352 11 sharedit sharedit PROPN bracis-19054 352 12 content content NOUN bracis-19054 352 13 - - PUNCT bracis-19054 352 14 sharing share VERB bracis-19054 352 15 initiative initiative NOUN bracis-19054 352 16 keywords keyword NOUN bracis-19054 352 17 fake fake ADJ bracis-19054 352 18 news news NOUN bracis-19054 352 19 one one NUM bracis-19054 352 20 - - PUNCT bracis-19054 352 21 class class NOUN bracis-19054 352 22 learning learning NOUN bracis-19054 352 23 positive positive ADJ bracis-19054 352 24 and and CCONJ bracis-19054 352 25 unlabeled unlabeled ADJ bracis-19054 352 26 learning learn VERB bracis-19054 352 27 transdutive transdutive ADJ bracis-19054 352 28 semi semi ADJ bracis-19054 352 29 - - ADJ bracis-19054 352 30 supervised supervised ADJ bracis-19054 352 31 learning learning NOUN bracis-19054 352 32 graph graph NOUN bracis-19054 352 33 - - PUNCT bracis-19054 352 34 based base VERB bracis-19054 352 35 learning learning NOUN bracis-19054 352 36 publish publish VERB bracis-19054 352 37 with with ADP bracis-19054 352 38 us us PROPN bracis-19054 352 39 policies policy NOUN bracis-19054 352 40 and and CCONJ bracis-19054 352 41 ethics ethic NOUN bracis-19054 352 42 profiles profile NOUN bracis-19054 352 43 ricardo ricardo PROPN bracis-19054 352 44 m. m. NOUN bracis-19054 352 45 marcacini marcacini NOUN bracis-19054 352 46 view view NOUN bracis-19054 352 47 author author NOUN bracis-19054 352 48 profile profile NOUN bracis-19054 352 49 search search NOUN bracis-19054 352 50 search search NOUN bracis-19054 352 51 by by ADP bracis-19054 352 52 keyword keyword NOUN bracis-19054 352 53 or or CCONJ bracis-19054 352 54 author author NOUN bracis-19054 352 55 search search NOUN bracis-19054 352 56 navigation navigation NOUN bracis-19054 352 57 find find VERB bracis-19054 352 58 a a DET bracis-19054 352 59 journal journal NOUN bracis-19054 352 60 publish publish VERB bracis-19054 352 61 with with ADP bracis-19054 352 62 us we PRON bracis-19054 352 63 track track VERB bracis-19054 352 64 your your PRON bracis-19054 352 65 research research NOUN bracis-19054 352 66 discover discover VERB bracis-19054 352 67 content content NOUN bracis-19054 352 68 journals journal NOUN bracis-19054 352 69 a a DET bracis-19054 352 70 - - PUNCT bracis-19054 352 71 z z NOUN bracis-19054 352 72 books book NOUN bracis-19054 352 73 a a DET bracis-19054 352 74 - - PUNCT bracis-19054 352 75 z z NOUN bracis-19054 352 76 publish publish NOUN bracis-19054 352 77 with with ADP bracis-19054 352 78 us us PROPN bracis-19054 352 79 journal journal PROPN bracis-19054 352 80 finder finder PROPN bracis-19054 352 81 publish publish VERB bracis-19054 352 82 your your PRON bracis-19054 352 83 research research NOUN bracis-19054 352 84 language language NOUN bracis-19054 352 85 editing edit VERB bracis-19054 352 86 open open ADJ bracis-19054 352 87 access access NOUN bracis-19054 352 88 publishing publishing NOUN bracis-19054 352 89 products product NOUN bracis-19054 352 90 and and CCONJ bracis-19054 352 91 services service NOUN bracis-19054 352 92 our our PRON bracis-19054 352 93 products product NOUN bracis-19054 352 94 librarians librarian VERB bracis-19054 352 95 societies society NOUN bracis-19054 352 96 partners partner NOUN bracis-19054 352 97 and and CCONJ bracis-19054 352 98 advertisers advertiser NOUN bracis-19054 352 99 our our PRON bracis-19054 352 100 brands brand NOUN bracis-19054 352 101 springer springer NOUN bracis-19054 352 102 nature nature PROPN bracis-19054 352 103 portfolio portfolio PROPN bracis-19054 352 104 bmc bmc PROPN bracis-19054 352 105 palgrave palgrave PROPN bracis-19054 352 106 macmillan macmillan PROPN bracis-19054 352 107 apress apress PROPN bracis-19054 352 108 discover discover VERB bracis-19054 352 109 your your PRON bracis-19054 352 110 privacy privacy NOUN bracis-19054 352 111 choices choice NOUN bracis-19054 352 112 / / SYM bracis-19054 352 113 manage manage NOUN bracis-19054 352 114 cookies cookie NOUN bracis-19054 352 115 your your PRON bracis-19054 352 116 us us PROPN bracis-19054 353 1 state state NOUN bracis-19054 353 2 privacy privacy NOUN bracis-19054 353 3 rights right NOUN bracis-19054 353 4 accessibility accessibility NOUN bracis-19054 353 5 statement statement NOUN bracis-19054 353 6 terms term NOUN bracis-19054 353 7 and and CCONJ bracis-19054 353 8 conditions condition NOUN bracis-19054 353 9 privacy privacy NOUN bracis-19054 353 10 policy policy NOUN bracis-19054 353 11 help help NOUN bracis-19054 353 12 and and CCONJ bracis-19054 353 13 support support VERB bracis-19054 353 14 legal legal ADJ bracis-19054 353 15 notice notice NOUN bracis-19054 353 16 cancel cancel VERB bracis-19054 353 17 contracts contract NOUN bracis-19054 353 18 here here ADV bracis-19054 353 19 129.74.145.123 129.74.145.123 NUM bracis-19054 353 20 hesburgh hesburgh PROPN bracis-19054 353 21 library library PROPN bracis-19054 353 22 er er INTJ bracis-19054 353 23 unit unit NOUN bracis-19054 353 24 ( ( PUNCT bracis-19054 353 25 3005732405 3005732405 NUM bracis-19054 353 26 ) ) PUNCT bracis-19054 353 27 northeast northeast ADJ bracis-19054 353 28 research research NOUN bracis-19054 353 29 libraries library NOUN bracis-19054 353 30 ( ( PUNCT bracis-19054 353 31 nerl nerl PROPN bracis-19054 353 32 ) ) PUNCT bracis-19054 353 33 ( ( PUNCT bracis-19054 353 34 8200828607 8200828607 NUM bracis-19054 353 35 ) ) PUNCT bracis-19054 353 36 nerl nerl VERB bracis-19054 353 37 ta ta X bracis-19054 353 38 account account NOUN bracis-19054 353 39 ( ( PUNCT bracis-19054 353 40 3006206169 3006206169 NUM bracis-19054 353 41 ) ) PUNCT bracis-19054 353 42 university university NOUN bracis-19054 353 43 of of ADP bracis-19054 353 44 notre notre PROPN bracis-19054 353 45 dame dame PROPN bracis-19054 353 46 hesburgh hesburgh PROPN bracis-19054 353 47 library library NOUN bracis-19054 353 48 ( ( PUNCT bracis-19054 353 49 3000184373 3000184373 NUM bracis-19054 353 50 ) ) PUNCT bracis-19054 354 1 © © ADP bracis-19054 354 2 2025 2025 NUM bracis-19054 354 3 springer springer NOUN bracis-19054 354 4 nature nature NOUN