The (Bene)fits of Compi l i n g a Special ized Database by Anne Wade Abstract The Centre for the Study of Classroom Processes (CSCP) at Concordia University in Montreal recently completed a three year project to compile a specialized database on the successful pedagogical approach of cooperative learning. Using the BiblioLinks software, the results of CD-ROM searches executed on a variety of source databases, were transferred into an in-house database. Utilizing the ProCite bibliographi c management software, a package designed to store bibliographic information, the database was customized according to the needs of the CSCP. The benefits of establishing a specialized database incl uded the abi l i ty to perform extensive searches on multidisciplinary sub-topics within the area of cooperative learning. However a number of problems occurred throughout the course of compiling the database, the majority of which related to limitations with the hardware and software. This article provides a chronology of those problems. Issues related to copyright and the publication of this database in the form of a comprehensive bibliography, are also addressed. The Centre for the Study of Classroom Processes (CSCP), is a research centre affiliated with the Department of Education at Concordia University, Montreal, devoted to the study of effec­ tive pedagogical techniques. Researchers at the CSCP have studied such topics as group composition (Abrami et aJ, in progress), dramatic play in early childhood education (Howe, et a!., in press), mental modeling (Schmid et al, in progress), the effectiveness of student ratings (Abrami et al, 1991), group size and microcomputer learning (Baron & Abrami, 1992) and coop­ erative learning strategies (Abrami et al., under review; Abrami et al., 1993). It is this latter subject area in which much of the CSCP research has been performed over the years. Because of this, researchers at the CSCP felt that it would be beneficial and cost-effective to establish a specialized in-house database of bibliographic information pertaining to cooperative learning and related areas. Bibliographic records would be retrieved from a variety of subject databases using the available online and CD­ Rom systems and subsequently transferred into one main data­ base. This database could then be used by CSCP researchers for in-house searches on sub-topics within cooperative learning. Bibliographies would be generated quickly and easily for the articles published at the CSCP and the database could also act as an online catalogue for the material housed within the CSCP Resource Centre. In line with the above, Williams (1986) discusses the ben­ efits of creating speciality databases due to the ease and availabil­ ity of inexpensive existing information technology and suggests: - 24 The ultimate user . . . is the customer of the user who has now become a producer of a speciality database. There is added value by virtue of some added processing, screening, manipulation, reworking of data to make the information more palatable to a given market. However, what was thought would take about six months to complete, actually took over three years and the concentrated effort of the part-time manager of the Resource Centre at the CSCP. Changes in the management, which occurred three times throughout the implementation of the project, contributed to the length of the project. What follows is a synopsis of some of the problems experienced and the issues involved in publishing the results of these efforts. These include: problems related to software selection, complexity of search strategies, limitations of both the hardware and software, lack of standardization between the CD-ROM databases, and copyright restrictions. Selecting the Software To begin with a package had to be selected which would meet the requirements of this project. Knowing that the literature searches would retrieve thousands of citations due to the vast quantity of information being published in the area of coopera­ tive learning, the software had to allow for the creation of a large database; it had to provide for the creation of user-defined workforms to input not only standard bibliographic information but also various unique fields specific to the CSCP; and it had to have the capability of transferring machine-readable records from a variety of online and CD-ROM databases . Two popular bibliographic management packages were compared: the Pro Cite Bibliographic System offered by Personal Bibliographic Soft­ ware Inc. (PBS), and Reference Manager by Research Informa­ tion Systems. After careful consideration, ProCite (Version 1 .4) was purchased as it was geared more towards library use. Various features included twenty-four "ready to use" or default workforms, each of which consisted of bibliographic fields specific to a particular type of material. For example, a workform existed for each of the following: journals, books, non-print Education Libraries • Vol 1 8, No. 1 material, reports, and dissertations (see Table I). ProCite also allowed the user to create up to six "user-defined" or customized workforms for those who wished to modify or add additional bibliographic fields and therefore did not want to use the default forms. Other noteworthy features included the ability to import records from remote databases (although a separate transferring package would have to be purchased for this), and its virtual unlimited record and database size. Conversely, Reference Man­ ager only provided workforms for book, journals and unpub­ lished works with no option to input new fields or to create your own work form. It seemed to be targeted more for the researcher. Because the CSCP also wanted to include its holdings, which consisted of many different types of material, Pro Cite seemed the logical choice. Two addi tiona! purchases were necessary before the project could get underway. A subscription to the online vendor Biblio­ graphic Retrieval Service (BRS) provided access to a number of relevant online bibliographic databases such as ERIC, Disserta­ tion Abstracts International and Psyclnfo. Finally, the purchase of the BiblioLinks software for BRS (also from PBS) was necessary as it formatted machine readable records retrieved from an online search into a format recognized by ProCite. Following these purchases, the information professional who had started the project, moved away and a second professional, who had experience using ProCite, was hired. Fol'mulating the Search Strategies Effective search strategies had to then be devised for each of the subject databases. It was critical that the strategies be broad enough to retrieve all relevant material with the smallest number of false hits. After numerous meetings with the researchers, a chart was devised which lis ted both relevant and unwanted terms and various sample searches were performed. Eventually a comprehensive search strategy was formulated which consisted of over 100 search statements tailored for the ERIC database that was offered online via BRS. Less extensive strategies were then devised for Psyclnfo, Sociofile, Dissertation Abstracts Interna­ tional (DAI), Education Library, Canadian Education Index, and the Ontario Education Resources information System (ONTERIS) databases. A ProCite database was created and customized workforms (or screens) were designed for each of the following types of bibliographic material: Books, journals, dissertations, reports, conference proceedings and non-print material. Fields were adapted within each of the workforms based on the projected needs of the CSCP. For example, fields were added to indicate: whether the i tern was located at the CSCP; ordering information; author affiliation and address; and other relevant information. The Advent of CD-ROM Technology Six months into the project there was, once again, a change in management and this author was hired to take over. CD-ROM technology had just arrived at Concordia, and three of the main Education Libraries • Vol 1 8, No. 1 Table 1 . ProCite's Workforms .a ·s: D I! .. ... D � � "' i ci ... T ll oo I � - · N c: l � :: "' ... .. .., 0 .. "' I! c: "" , .., ,!, • � :J A. � C ;! � ! ...; ... ob c: D ... l i • ... c: ..c: .!l .; • .. t "" i! ..c: u " .. ... !i l:l ... . , n " "' t:, -;; !l Z Z :ii .!! i .!! "t :J - - :s • �o�t o .a « ... D. a. Q - Z C: "" ..c: ... • � .: .., : II . .. U Q � -r- � i ., ... ;c "' . ��--­ � � � ; i. l .. .. D .c "' I "" D .:! ... ... �---� !l ... . .. • ! i -1- i � .. ... � r--� 1: i .. .. :0: " • � ....__....=.._ source databases (ERIC, PsycLit, and SocioFile) that were to be used on this project were purchased by the university from SilverP!atter. Because searching on CD-ROM was less expen­ sive than online (Concordia charged $5/hour vs BRS's connect time charges) it was decided that CD-ROM searches would be performed whenever possible. Thus a second transferring pack­ age, BiblioLinks for SilverPlatter, was purchased from PBS. The next couple of weeks involved testing the CD-ROM hardware and adapting the search strategies to the SilverPlatter software. When it was time to run the searches, initial problems related to both hardware and software constraints on the CD­ ROM systems: SilverPlatter would not accept more than 100 search statements; search strategies could not be saved for quarterly updates of search results; and the complexity of the strategies - long search statements and the combination of numerous terms (e.g., SMALL nearl (TEAM* OR GROUP*) near2 (LEARN* or INSTRUCTION or WORK*) would jam the hardware. This was particularly frustrating as the computer would seize up, usually around the 90th search statement, thus requiring the re-input of the entire search strategy. The search strategy could not be sectioned into smaller units as the state­ ments were all interrelated and had to be executed in one session. Also, because the retrieved sets were so large, the computer would jam after having downloaded 590 of the 600 records retrieved, with the only option being to reboot the computer and re-enter the entire search session. The majority of these problems occurred as SilverPlatter was refining its software and Concordia 25 - was upgrading its hardware. Thus, the project had to be put on hold a number of times in order to wait for the respective hardware/software upgrades. Months later the searches were implemented successfully. BiblioLinks and the Transferring of Records into ProCite The results of all the CD-ROM searches were saved as ASCII files on floppy disks. Before transferring the records into a ProCite file it was necessary to retrieve the ASCII files in WordPerfect for what should have been minimal editing (e.g., adding a notation at the beginning of the file stating the name of the database records were retrieved from). When transferring bibliographic records into a database management package, it is critical that the fields correspond so that the information is sent into the appropriate fields. This can be tricky as the source databases, such as ERIC and PsycLit, all contain different fields and/or field labels; some abbreviate the field label, while others do not. As well, each database uses a different bibliographic style. For example, one index may use the format "Online, v33, no. 1, 1992" while another uses "Online, vol. 33(1), May 1993". Even within the same database the source field can change. For instance, PsycLit records changed their format in 1977. The advantage of a transferring package such as BiblioLinks is that it is designed to recognize such differences and little, if any, editing on the ASCII files should be necessary. BiblioLinks offered separate software packages for each of the various online systems. However, due to the fact that CD­ ROM technology was so new, PBS did not have packages for many ofthe different CD-ROM systems such as those offered by, University Microform International (UMI), Micro media and the Institute of Scientific Information (lSI). Therefore, for those CD­ ROM searches performed on UMI's Dissertation Abstracts Inter­ national and Micromedia's Canadian Education Index, macros had to be developed within WordPerfect and subsequently run on the ASCII files in an attempt to adapt the records into a format similar to that of SilverPlatter. Various field labels had to be abbreviated, renamed, and re-assigned for each of the different databases in order for them to be recognized by the Silver Platter BiblioLink. However, even the ASCII files from the Silver Platter searches had to be edited for successful data transfer and fields had to be re-assigned within BiblioLinks (fortunately a relatively straight­ forward task). Some of this reformatting was due to the fact that user-defined workforrns had been developed within ProCite. In retrospect, it would have saved a lot of time if ProCite's default workforms had been used and perhaps adapted to the needs of the CSCP. Once the ASCII files were formatted appropriately, they were ready to be uploaded into the ProCite database. However, there was more editing required withinProCite! Because ProCHe's Version 1 .4 did not allow for global changes, each record had to be edited individually. For example, journal fields were not always parsed properly, and information had to be moved into the appropriate fields. Also, hard carriage returns that were inserted - 28 at the end of each line in the abstract field had to be deleted. ProCite has some word processing features, such as Block and Move, thus making editing a little easier, but it was still time­ consuming. One major problem occurred with the ERIC database, which indexes journals, reports, conference proceedings, dissertations and books. Each different type of bibliographic material should be transferred into the corresponding workform (e.g., journal articles should transfer into the Journal workform, reports into the Report workform etc.). However, BiblioLinks is not able to distinguish between different bibliographic records within the same ASCII file and all records were dumped into the Journals workform. Within BiblioLinks there is a feature which can be changed prior to transferring the records. This feature allows one to re-assign a database, such as ERIC, so that records are transferred into a workform other than the default. What this meant, however, was that individual macros had to be set up within WordPerfect and run in the ERIC ASCII file to first move all the books, then the reports, then the conference proceedings and finally the dissertations, into their own respective ASCII files. Each file was then subsequently transferred into ProCite once the BiblioLinks command had been changed to reflect the type of bibliographic material that would be transferred from the ASCII file. Unfortunately the macros were not fool-proof and it was easier to have the journals in one file and all the other material in another. BiblioLinks was customized to transfer records in the "other material" file into the Reports workform. Workforrns for records which were not reports were then changed to the appro­ priate workform within ProCi te. Once the workforrn was changed, some information had to be moved into the appropriate field. It had been decided that each record would be assigned a code which noted the database from which it had been retrieved. For example, ERIC records were assigned CLl, PsycLit records CL2, SocioFile records CL3 etc. These codes had to be assigned to each record individually as there was no feature in Version 1 .4 which globally copied information to a field. In addition, the duplicate detection command was run after records were trans­ ferred from each database to eliminate all duplicate records. It had been determined that priority would be given to ERIC records and next to PsycLit records as the descriptors assigned to ERIC records would be more relevant. If a record was picked up on two or more searches, it was coded to reflect this. Another task was to flag those records which could be found at the CSCP, because the CSCP has an extensive collection of journal articles which has been collected over the years for various research projects. It was expected that all the articles on cooperative learning that were housed at the CSCP would be retrieved on the searches and it would be useful to flag these. A Research Assistant was hired to input into a ProCite file, the author and title of all the articles housed at the CSCP. This file was then merged with the main database and the duplicate detection feature was subsequently run. All those records which were retrieved on a search and located at the CSCP were then coded to reflect this fact and the skeleton record was deleted. Any remaining articles that were at the CSCP but had not been picked up on a search were subsequently edited and the remaining bibliographic information was manually entered. Education Libraries • Vol 1 8, No.. 1 PBS released Pro Cite 's Version 2.0 during the editing phase of the project and it offered both a Global Search and Replace and a Global Cqpying command which saved considerable time. But even with the improvements in Version 2.0, some of the changes, most ofwhich related to capitalization, had to be done one record at a time. For example, contrary to the American Psychological Association (AP A) standards, ERIC capitalizes the first charac­ ter of each word in a title; some of the databases use full capitalization for index terms, others do not; and UMI's DAI uses full capitalization for their entire record including the abstract. It should also be mentioned that because PBS did not release an upgrade to BiblioLinks (Ver.l .1) along with their upgrade to ProCite, ASCII files had to be transferred into a database using the old ProCite (Ver.1 .4) and then converted to the new version of ProCite. Downloading the Results into WordPerfect After many months of time consuming editing, and the need to upgrade the hardware to a 486, SOMhz computer as the database was growing at an alarming rate, the results were beginning to reach fruition. Subject searches were run for re­ searchers and graduate students at the CSCP. They were suffi­ ciently impressed and the results of these searches were downloaded into WordPerfect to be printed as subject bibliogra­ phies. Still more editing required within WordPerfect. Because ProCHe does not move the title intp the author field for anony­ mous articles, the date had to be moved. Also bold and underlined characters would not transfer into the WordPerfect file, so additional macros had to be set up. Once again, some of this editing was reduced with Version 2.0 as it provided the ability to save files in WordPerfect format and ASCII characters were downloaded correctly. Needless to say, a lot of time would have been saved if this new version had been available at the start of the project. Sharing the Results After three years, the CSCP's specialized database on coop­ erative learning was finally up and running. Numerous printed sources, including: the Education Index, British Education In­ dex, Educational Technology Index, Books in Print, publisher's catalogues, and various directories to audio-visual material were scanned and all relevant material was manually entered into the database. The final product consisted of over 10,000 citations to books, journal articles, dissertations, reports, conference pro­ ceedings, curriculum and non-print material. Each record con­ sisted of a full citation with abstract, keywords and the availabil­ ity of the document (when provided). Updates of the manual and CD-ROM searches were implemented as the sources were up­ dated to ensure currency of the database. Numerous in-house searches were performed relating to specific areas within coop­ erative/collaborative learning, such as: multiethnic students, early childhood education, second language instruction, compu­ ter-assisted instruction and disabled students. With Version 2.0, global changes could be made to selected records or to the entire Education Libraries • Vol 1 8, No. 1 database; Pro Cite search strategies could be saved directly within the program; even the search results could be saved; and compre­ hensive searches could be run relatively quickly with the Quick Search feature. Results were so satisfying that it was decided to share them with others outside the university. There were two options -the CSCP could publish a bibliography on cooperative learning in printed format, or the database could be offered to the public in machine-readable form (on diskette or CD-ROM). It was decided that both options would be pursued as different markets would be targeted for each of the two products. Having downloaded such a vast quantity of information, it was unclear as to whether the CSCP would be within the copyright boundaries in publishing the results, especially due to the fact that the abstracts were copyrighted. The end-user is unaware of how one can use the information retrieved on a CD­ ROM search. This is due to the fact that they do not know the terms within the licensing agreement between the institution, in this case, Concordia libraries, and a database supplier, such as the American Psychological Association. This is a common problem within libraries and Duggan (1991) suggests that signs should be posted next to the CD-ROM workstations, much like those next to the photocopy machines, which state the copyright restrictions as outlined in the licensing agreements. This is an excellent idea but one that is not being practised in mo�t places at the moment. Because of this uncertainty, requests had to be sent out to all the various database suppliers - could the specialized database be published in printed form? "No problem," said the Educa­ tional Resources Information Center (ERIC), "bibliographic data has been generated with the support of tax dollars and is in the public domain and may be used without restriction." In fact, ERIC's objective is to achieve widespread dissemination of its data at the lowest possible cost and they encourage projects such as this. University Microfilms Inc. said there was no problem as long as it was stated that the full text of the DAI records were available from UMI. The American Psychological Association was not quite as forthcoming. They required written authoriza­ tion for replication of PsycLit records in a printed bibliography. Similarly, the International Sociological Association would give permission only after receiving a complete list of accession numbers for the Sociofile records that would be used. It was felt that the above requirements could be met and samples of the database were then sent out to a variety of American, Canadian and British publishers in an attempt to find a publisher for the bibliography. The general consensus was that bibliographic information lends itself to electronic publishing and the printed format was passe. Only one publisher expressed an interest. However, they required camera-ready copy and would provide no funding to pay for this . They also required that the CSCP would agree to purchase 100 copies of the final product. The idea of a printed comprehensive bibliography on coop­ erative learning began to fade. Efforts were then focused on finding a way to offer the database in machine-readable form. Fortunately, the Ontario Institute for Studies in Education (OISE) conveyed an interest in offering our database on CD-ROM along with their ONTERIS database. The database suppliers were then approached once again to find out their copyright restrictions for 27 - publishing the database in electronic form. The copyright bound­ ary had been hit. Although both ERIC and UMI were agreeable, no other commercial database supplier would give permission to publish these results in electronic form. Interestingly, UMI requested a letter which outlined the entire project. This was reviewed by a committee which reached an affirmative decision weeks later. Our request was obviously scrutinized and given careful consideration. UMI must have realized that our desire to publish a bibliography in electronic form represented the tip of the iceberg in what would seem to be a rapidly expanding electronic revolution. What Does the Future Hold for Specialized Databases? Having managed the last two years of this project on! y to find at the very end that it would be difficult to share the results with the public, was extremely disappointing. But the quest continues. The CSCP has recently approached various funding agencies in an attempt to obtain a publications grant. Provided one of these proposals is approved, the publisher who expressed an interest, but required camera-ready copy, will publish the comprehensive bibliography in the print format. In the meantime, CSCP mem­ bers, educators, researchers and graduate students have certainly benefitted from the numerous in-house searches which have been performed and which will continue to be performed into the future. Additionally, the CSCP has developed a partnership 'Yith the International Association for the Study of Cooperation in Education (IASCE) in which customized bibliographies are offered to IASCE members- this is allowed as long as i t is a non­ profit venture. In fact the CSCP is presently considering compil­ ing another specialized database on post-secondary education. Many ofthe problems that were experienced throughout this project would not exist to the same extent now due to advances in the software - bibliographic management systems, transfer­ ring packages (e.g., PBS is presently working on an upgrade for BiblioLinks), and CD-ROM systems have all been improved over the last couple of years. In May, PBS came out with Pro Cite Version 2.1 which corrected ProCite's serious problem of not recognizing extended memory. Because of this problem, the user would receive an error message when trying to display "pop-up menus" and authority lists within a large database, even though , sufficient extended memory was present. Other problems would occur when editing or deleting a record -Pro Cite would prompt you to save the record immediately as the computer was running out of memory. With Version 2.1, ProCite now recognizes up to 4Mb of available expanded memory - but due to ProCite's programming language it is still not able to recognize extended memory. Version 2.1 also fine-tuned other less critical tasks such as the Repair, Quick Search, and Duplicate Detection commands. It is expected that PBS will continue to provide improvements to its software and will develop additional BiblioLinks packages for all the various CD-ROM systems. More significantly, we are beginning to see systems such as Innovative Interface's InnoPac provide the option of downloading records in ProCite-specific export format thus eliminating the - 28 need for a transferring package such as BiblioLinks. Similarly, more powerful hardware has become affordable-extensive searches can be performed much more rapidly. That is not to say, however, that the compilation of a specialized database will be effortless as lack of standardization among the systems will always present problems. What is of more concern is the issue of copyright and how one can use the information retrieved from these remote databases. Without a doubt, the transferring of machine-readable records into in-house databases will be a trend of the 1990s and beyond. Increasing access to worldwide library collections, full text journals, and bibliographic databases through systems such as Internet, coupled with the advances in the storage of electronic information such as CD-ROM, will invite the creation and maintenance of specialized databases. The thought that these databases can only then be used internally seems inconceivable, especially knowing the amount of time and effort it can take to create such a beast. The issue of"who owns the data?" will have to be addressed sooner or later. References Abrami, P.C., & d'Apollonia, S. (1991). Multidimensional students' evaluations of teaching effectiveness-The general izabil i ty of "N=1" research: Comment on Marsh. Journal of Educational Psychology, 83, 4 1 1 -415. Abrami, P.C., Chambers, B., Poulsen, C., Spence, J. & Lou, Y. (in progress). The effects of group composition on achievement: A meta analysis. Abrami, P.C., Chambers, B. , Poulsen, C., Howden, J., Kastelorizios, K., Wagner, D. , & Glashan, A. (1993). Using cooperative learning. Dubuque, IA: W.C. Brown-Benchmark Publications. Abrami, P.C, Chambers, B. , Poulsen, C., Kquros, C., Farrell, M . , & d' Apollonia, S. (under review). Cooperative learning and social support: A primary i nvestigation and research i ntegration. Journal of Social Psychology. Baron, L.J., & Abrami, P.C. (1992). The effects of group size and exposure time on microcomputer learning. Computers in Human Behavior, 8, 353-365. Duggan, M.K. (1991). Copyright of electronic i nformation: Issues and questions. Online, 15, 20-26. Howe, H., Moller, L., Chambers, B. , & Petralos, H. (in press). The ecology of dramatic play centres on children 's social and cognitive play. Early Childhood Education Research Quarterly. Schmid, R.F., Abrami, P.C., d' Apollonia, S., DeSimone, C., & Poulsen, C. (in progress). Mental modeling theory applied to the cognitive analysis of learning strategies and tactics. Williams, M. (1986). Transparent information systems through gate­ ways, front-ends, intermediaries and interfaces. Journal of the American Society for Information Science, 37, 201-214. Author's note: ProCite for the IBM or Macintosh can be purchased for $395 U.S. directly from PBS Inc., P.O. Box 4250, Ann Arbor, MI 48106 (Tel : 313-996-1580, Fax: 3 13-996-4672) or from PBS 's Canadian distributor, M.E. Phipps, 143 Norfolk St., Guelph, Ont. (Tel: 5 19-836-9328, Fax: 5 19-836-2623). PBS has recently devel­ oped a network version of ProCite. The Library BiblioLinks and Online BiblioLinks are also available for $295. BiblioLinks pack­ ages are avai lable for D ialog, B RS, OCLC, Si lverPl atter, Education Libraries • Vol 1 8, No. 1 WilsonDisc, and various other online services. PBS software is also avai lable at various university bookstores at discounted prices. The author would l ike to acknowledge the sound advice provided throughout this project by Dr. Philip Abrami, Jacqueline Dealy, and Catherine Poulsen. Anne Wade is Manager of the Resource Centre at the Centre for the Study of Classroom Processes (CSCP) , Concordia University and also a part-time Reference Librarian at Westmount Public Library, Montreal, Quebec, Canada. Anne Wade, CSCP, LB-58 1 , 1 455 de Maisonneuve Blvd W . , Montreal , Que H 3 H 1 Y5, Tel : (51 4) 848-2020, Fax: (5 1 4) 848- 4520, E-Mai l : Wada@vax2.concordia.ca AB@LA proudly announces the publication of . . . Un gu ide d'excel lents l ivres fran9ais pour enfants et adolescents This is an annotated l ist of almost 1 50 titles of board books, nursery rhymes, easy readers, picture books, junior and young adult fiction, and non-fiction. Carefully selected by a committee of librarians, these books were all published between 1 983 and 1 993 . Use this guide as a giveaway to your favourite borrowers and customers, to get ideas for birthdays or hol iday presents for the special chi ldren and teenagers in your l ife. Let Lire! help you with book selection and as a buying guide. Price: ( 1 -5 booklets) $4 each; (6-1 0 booklets) $3 .50 each; ( 1 1 + booklets) $3 each. (plus 50¢ handling and 86¢ postage per booklet ordered) For further information, please cal l : Marie Eberlin, AB®LA Secretary, at (5 1 4) 630-4875 Send cheque or money order to : QU EBEC LIBRARY ASSOCIATION, LIRE! P.O. Box 1 095, Pointe-Claire, QC, Canada H9S 4H9 Education Libraries • Vol 1 8, No. 1 29 -