Acta Polytechnica Vol. 43 No. 412009 To Whom Belongs Conceptual Design? J. Bfla Th.e lieLd of Conceptual lytlry is aery alfue and is rapidly deteloping. This paper inaestigates the disciplines and domains which sltlsta2llaUy lonn its profilc. Tlwre are considered disciplines such as Semiotiis,'Fornnl Ligic, Euolutioiary analogies, euali,tativeMod.elltng, Ontologies, Artifitial Intelligence and Ernergent Syntlrcsis. The answer to the quistion posed. in ihe title'iies'niwadays in disciplines rel"ated to Cogniti,ae Science. Keyw ords : c onceptual de sign, semiotics, ontologies, U M L, artificial intelligenc e. I Introduction ConceptuaL Design. remains a very artractive field of re- search. There are two reasons for this: a free space for model- ling of creativity, and the opportunity to apply novel means of Artificial Intelligence. Preparing this papeq, use was made of information from many sources of Conceptual Design Support and Conceptual Design Process (CDP) modelling (and we apologise to ail which are nor introduced in the References). 'fhe essence of CDP near ro the context of this paper is available, e.g., in 13, 4, 5, lg]. Attempts at constructing a d,eeper fonnal dcscription of CD P w ere done, e. g., in I I 3, 20]. The means of aerifuation of CDP results were suggested, e.g., in U, 2, 71. (In I I I ] there were discussed means of verification of SW products for designin thz pre-irnpLemmtation phase). Support, for coweptual dzsign of artefacts and ways for modclling crcatiuily in CDPwere presenred, e.g., in [3,9, 10, 13,28,29,35]. The application of structural ann fi sis mz thods in CDP were used, e. g., in [8] (Yourdon Structural Analysis) and in [21] (UML), [20] (OMI-UML). Special computer CDP support systazs were pre- sented, e.9., in 13,6,7,8, 12, 15,241. 2 Semiotics, formal logic and evolutionary models The outcome of the process of Conceptual Design (in technological fields) is usually undersrood tobe a scheme. Thz schune has substantial features ofa product or system which is designed but need not necessarily contain geometrical and quantitative data. Many Conceptual Design activities may be studied in the field of Semiotics, in a small fieldwhich is inves- tigating the specific cognitive relations berween the signed and the signing, and their reflective and cognitive functions. Comparing figures from Leonardo da Vinci's "notebook" with Olesen's figures from [9], we have to admit thar rhe tech- niques of conceptual reasoning and of expressing conceptual ideas and their essence have not changed fundamentally in a period of 400 years. As conceprual designs there might be considered not only schemes of Leonardo's submarines but also the schemes of Diirer's figural compositions [14], the schemes of Michelangelo's constnrctions and also the scheme of Alessandro Marcello's (1686-1739) oboe concerto. In all these examples, the scheme lives within a leael of conceptua,li- sation which drives the understanding of the scheme, What is the purpose of the schcme? Tfu schemc presents the form of the designed system and explains its function. How deep this prcsentation and these explanations are depends on the on- tology r.vithin the framework of which the schemz rvas formed. Formal logic has tried to help in many aspects of concep- tual design theory. This paper menrions only two: o the formation and processing of concepts (Frege, Tichli, Materna), o lvays of transferring truth (Godel, Gentzen, Robinson). The first line of research established concepts such as specific structures, which are composed in conceptual con- structions. The second line of research discovered the fact that the deeper semantics and pragmatics of conceptual constmctions may be described by a formal logic system, by a system which is able to conrrol what is possible, what is impossible and what is correct. (Using such a system we can stipulate that legs are not parr of rhe head and thar thewheels are not situated on the roofs ofcars). In conclusion, the contribution of Semiotics and Formal Logic to Conceptual Design is natural and nor too sophisti- cated. Semiotics tries to grasp the process that takes place benveen the model and the brain (mind) of rhe designer dur- ing designing, but its formal means were adapted to a static investigation of language rather than design issues. The auro- matic mechanisms which assign sign formations to ideas during language phenomena are unfortunately not visible by the formal means of Semiotics. Formal Logic, howeve4 helps Semiotics, but post factum., Hypotheses investigating the "paths" of design ideas which are not necessarily mediated through lnngrnge are fresh and alive nowadays. In this context, the research lines of BioSemiotics [9] and Evolutionary Analogies [33] which in- duce images about a more natural formation of artefacts are very interesting. Hg. I illustrates the interaction benveen Niche Space and Design Space, as inroduced by Sloman in his "speculation" about Evolution [33]. Fig. l: Interaction between Niche Space and Design Space Acta Polvtechnica Vol. 43 No. 412003 Niches are the carriers of "requirements" in living ele- ments which "may produce prcssure for evolu.tionary change". Howeveq they are more than static lists of require- ments. They also own reflective functions and a very useful complex of properties which Sloman called affordnnce (the ability to decide the actual role (of the niche)). In other words, affordance induces variants of functions. Though Sloman's paper [33] considers real long time evolution, the analogy for CDP is very attractive, and an especially important feature is the process between the Niche Space and the Design Space, which is a process between the signed, and the signzng, a meta- phorical image of a special co-evolution: "Possible duigns and possible nirhes are linked fu desuiptions of ways in which differmt dzsigns mntch a pafihul.ar ni.che and tht samz design ruztches dffir- mt nfuhes. Since mismatches can produce pres$ures for changes in dcsigns, and this can produce neu niches, Ieadmg to new kind of mabhes and, misruttrhes, ue hnae interacting systems concurrently tracing trajectories through duign space and through niche space with complex interacting feedback loops" , [33]. Sloman differenti- ated three essential types of feedback loops (Fig. 1): iJoops (individual learning and developm ent), eJoops (evolutionary development) and r-loops (repair loops: an external agent replaces, repairs or adds a new feature. It may then jump to a new part ofdesign space and niche space.) The llagment of an evolutionary analogy introduced above contains mainly novel images and terms, it turns our attention to a process the goal of which is different from 'lcorrect understanding" of signs. In otherwords the commu- nication aspects play a less important role (in the process illustrated in Fig. l) than in the classical fiamework of Semiotics. 3 Concepts and qualitative modelling This section will sketch a small "ontology" for work with concepts.and related categories. After explaining that con- ceptual design is based on operations with intentions rather than with concepts, a short reasoning about a calculus for Conceptual Space will be presented. First, it is necessary to differentiate between concepts and 'intzntion"s. Coneepts are abstract ideal categories, and according to their use they belong to knowledge. From the procedural point of view they are identifuation procedtnes which identifi objects. (Objects are used here as entities outside the subject. There is no relation here to the Object Oriented approach.) Each concept has gxpressioR, con;nt (substrate), structure (according to Bolzano) and meaning. A typical example of a concept is: "Primes" [36]. If we know this concept, we also know the procedure for identifying possible numbers as primes or not primes. What is important is that this proce- dure is our internal knowledge, we do not need any empirical facilities or external assistance. Il for the identification pr oce- dure, we need some empirical facilities and operations, e.g., for identif ing the situation "actual temperature in block A', we speak aboutem,pirfual concepfs. Howeveq the rcsults of iden- tifring empirical concepts are not objects butintentians. Intentions (as introduced, e.g., in [26, 36] ) represent roles which may be played by objects. (E.g., 'lto be a support for", "to be an engine of", ...). Intmtions are mappings from time-space states of possible worlds into a space of values (and they have no internal structure in general). It is important to emphasise that whilst the definition domain of these map- pings is the same (time-space states of possible worlds, but depending of course on the actual instance of time-space states in a possible world), there are four Dasfc spaces ofvalues: 'rspace of truth values", "space of individuals", "space of classes of individuals" and "space of numbers". Correspond- ing with these value spaces there are four basic intentions: propositions, ffices, propertezs and quantities. After this small excursion into the background of conceptual constructions it is not surprising that the categories by which we operate in Conceptual Design are more intentions than concepts. trig.2 introduces a simple image of the evolution of a Conceptual Space (a space with, concepts and intentions) and a Calculus (in our case a calculus for modelling and sup- porting CDP). A mental image from the external world is not necessarily the principal category in this scheme. Its main task is to navigate attention during the structuralisa- tion of Conceptual Space. Of great importance are Semiotic Activities, which execute the relations between concepts and intentions and control their evolution. Experience with vari- ous formal means has shown that though there are some general relations between Conceptual Space and Calculus, there are strong limitations in the development methodol- ogies, which must be avoided with the help of add:itional donnin knouledge. Empirical experience with operations involving inten- tions and concepts leads to qualitative objects, and as a Mental image of external world \ Conceptual Space <-[- --*----l I catcutus I---->l I Fig. 2: Conceptual Space and Conceptual Calculus 4 Acta Polytechnica Vol. 43 No. 412003 consequence to qualitative operations. It is interesting to consider the nature of such operations (returning to the issue of calculus (from Fig.2)). (In order to avoid the necessity to specifi when we are working with intentions and when with conceprs (for each case) we will speak about members (x, y, z, ...) of Conceptual Space (CS). These members do nor represenr .*p..rrio.,r, contents, structures, ... of concepts, nor values of intentions, but they represent their special semanric content. They repre- sent all consequences which are relevant to the specification of a Conceptual Design problem and rvhich can be derived from the considered members.) Some basic implications are available even in a very weak structure of Conceptual Space where E is a carrier of CS, u is a binary operarion of CS mem- ber synthesis and - is a relarion of 'st.ong similarity which is considered as a relation of tolerance. (For "not -" we use symbol "4".)The follorving axioms corresponds to empirical exPenence: VxeE,(xux)-x, ((:*,y€E)AND(x +v))-(("uy)+(yux)), (A2) ((t *,y,2 e E)AND((w /- y)AND(x + ')))= (A3) = ((* uy) + (x u z)). From these axioms we can derive the following theorems: Tl: Operation u is nor associative (with regard to relation -): ((: x,y,z e e)eNo((x + y)aNo(y + ')))- ')= ((* u (y u z)) + ((x u y) '4)) T2: Operation u is nor bisymmetrical (with regard to rela- tion -): 3 w, x,y, z € q(((w u x) u (y u z)) + ((rv u y) u 1" u z))). t2l As consequences ofAl -T2 we can find that operation u is neither additiue nor rnetrfu (in terminology introduced in, e.g., [25]). This implies thar for Conceptual Space there is no general method fbr constructing an apprcpriate metric func- tion for measuring the "distance" benveen the members of CS. And fiom this ensues that therc is no seneral method for designing the feedback control loop needed for programma- ble development (evolution) of CS members. The above thoughts about the nature of a Calculus for Conceptual Space can be under.stood as avery small contribu- tion to the discussion of why the operations in Conceptual Design are rather qualitative, why quantitative methods in control of CDP are unrratural, and why compurer support for CDP needs special approaches and means. ' 4 Ontologies and conceptual structures Research on Ontologies nowadays belongs in rhe field of Artificial Intelligence, but it also plays a significant role in Conceptual Design. The term ontology has been used in the follorving senses [30, 3 I ]: o a philosophical approach to the investigation of"being", o an informal conceptual system, r a formal semantic account, r a specification of a conceptualisation, . a representation of a conceptual system via a logical theory, o a vocabulary used by a logical theory, o a metalevel specification of a logical theory. For this paper the most convenient interpretation is "Ontology is a specification of a conceptualisarion". From the knowledge representation point of view, ontologies are semantic net\4'orks, very appropriate for c,rnceptual model- ling. The main objectives pursued by research on ontologies are: r sharing and interchange of knol'ledge, . management of knorvledge, o data retrieval. It is clear that ontologies parricipate in all conceptual de- signing and their main merir is the opportunity to combine different professions, expert knowledge and points of view in the rcquired domain. Research on ontologies is at present focused on representation and on semantic modelling. In addition to Ontolingua [34] there are a few serious candidates for this place. A promising candidate seems to be a combina- Cs=(E,u,-), (Al) (Al) Building constructlons Fig. 3: Example of a hierarchical structure of ontologies Acta Polytechnica Vol. 43 No. 412002 E$,Material: TMat 6HPlaces:field SCutTool $ToolBody Fig. 4: Fragment of a multi-view ontology of a product tion of OMT methodology (Object Modelling Technique) [6] and the UML language (Unified Modelling Language) [17]. UML is able to represent semantic nerworks and OMT is important for developing a factual ontology. Iig. 3 shows various types of ontologies neighbouring on engineering ontologies. Fig. 4 shows a fragment of the ontology [32] of a product description (combining the points of views of the product designeq, technologist and CNC software engineer) developed in UML. 5 Computer support for CDP From a purely white-collar point of vieq CDP could go on some database in a network of retrieval and composition procedures. Howeve5 this is a rather administrative image. CDR differ in their internal strucrure and the compurer support for their components also varies. The analysis of conceptual design processes underlines the following crireria by which various CDR can be compared with each orher by means of their most important characteristics: Pl. Translation of the initial specification into CDP. P2, Decomposition of Functions and Structures. P3. Proper method for forming conceptual constructions. 6 P4. Verification of the correctness of the CDP result function. P5. Way of modelling the emergence of novel solution. In order to concentrate information about the level of CDh and about the type of designed systems, thewhole field of various CDR will be considered decomposed into three classes of Conceptual Designs: Al Corueptml d,uign of Configuratioru (flats, buildings, parks, allocation of machines in halls, ...). B) Curcephnl dcsign of technological companents, machines ard, dzaires (holders, attachment tools, frames, bicycles, cars, paragliding sets, refiigerators, heat pumps). C') Concephnl dzsign of systems (control systems, technological systems, transport systems, telecommunication systems). Note 5.1: The above decomposition is one of many. It is condi- tioned by criteria Pl, .. ., P5 and by a certain temporary inter- est of designers. It is an example of a decomposition, and it could not induce a discussion of the rype "Is the paragliding set a machine ?". A, B, C, represent certain ontologies as lev- els of conceptualisation. (It is clear that a refrigerator may be considered as an element in all classes,\ B, C, according to need.) The orientation results of the evaluation of classes A-C by criteria Pl-P5 are as follows: Acta Polytechnica Vol. 43 No. 4/2003 Conceptun l design of Confgurations PIA: The translation of the specificarion is performed by a graphic interface. The programming environment may be of the Prolog type. PZA: Decomposition of the srrucrures is determined by the content of the library of structural elements and parts. P3A: The composition operarions for elernents and parts are defined. P4A: Direct verification of the specification without the need for any additional computation. P5A: Visual interaction of shapes - occasional emergence. (The Magic of M.C. Escher, [37].) Note 5.2: If we consider the design of configurations of rna- chines in a hall the conceptual know how is the order and arrangements of the machines, nor rhe solvabiliry and possi- ble productivity of the designed allocation ser. If rve need simulations the problem belongs to class C. Conreptual duign of principles in technological component; machines and deuices PIB: Graphic interface. The specificarion contains static, kinematic and dynamic paramerers of the designed target. We assume the translation of the specifica- tion into some available programming language (e.g., Delphi, C++, Prolog). P2B: Decomposition of the functions is a substantial task. It is necessary to define the funcrions of the components. Decomposition of the sfuctures is easy. We assume a developed database (library) of design elements and parts. P3B: Composition operations which realise a design as the composition of principles, functions and structures of elements and parts into wholes are performed by rules of composition and restrictions. We assume a devel- oped rule-based knowledge base. P4B: We take into consideration small verification compu- tations or small simulation experiments. P5B: The modelling of the emergence of a novel solution is not assumed. Concephnl fusign of systems PIC: The translation of the specificarion is performed by a graph-symbolical interface. The specificarion con- tains behavioural description, many functional and structural parameters of components and wholes. The application of special methodologies is assumed (e.g., OMT) and translation of the specificarion into special languages and program environments (e.g., UMI-, STEP 7). P2C: The decomposition of the complex function of the designed system into sub-functions is performed by special rules and by means of the database of ele- ments and subsystems. There are supposed a devel- oped knowledge base and a library of elements and sub systems. P3C: The proper method for forming conceptual structures only extends (by rules for compositions and restric- tions) the decomposition process from P3B. P4C: No detailed verification of the functional correcmess of the system is available within the framework of CDp. This would require large simulation models and many experiments. There are three ways to approach this important issue: o by approximating the behaviour and the properties of the developed prototypes. by using some novel method of verifrcation without experiments. by utilising UML represenrarion of the system and by means of some appropriate CASE system ro gen- erate the code of simplified simularion and visuali- sailon programs. P5C: Modelling the emergence of a novel solution is not assumed. 5 Artificial intelligence in conceptual design This section rvill introduce examples of AI sysrems devel- oped to support CDP A. "Stani"ard" AI CDP Sztpport Slstems This class includes systems which use decomposition and composition operations in the Function-Structure platform and rvhich rvork rvith a pre-formed dedicated database and knonledge base. Examples of such systems are Galileo [6], AIDA [8]and GPAL [5]. GALILEO is a knorviedge-based CDP support system. [6] presents examples of its application for conceptual design of *r'o classes of devices (which were commercially interesting) but rve can imagine how difficult (or how easy) it would be ro develop its application for the conceptual design of an an- other similar system. The kernel of the system is a knowledge base rvhich contains atomic and partial essential stmctures of the type "Required function + lvleans of its rcalisation ". The principle of CDP lies in decomposing the global function of the designed system (included in the specification) into subfunctions - Fig. 5. The verification process takes place during designing, and its efficiency is limited by the content of the library of elements and parts and by the implemented constraints. AIDA is described in [8] as an AI system for computer sup- port for Conceptual Design of complex systems. It is based on a combination of three AI tools: Case Based Reasoning ficr suggesting the initial proposals, Rule-Based Reason- ing to assess these proposals and their functional qualities (i.e., small computations and checking points), and Con- straint-Based Geometrical Modelling for visualisation of the proposals. These tools are developed as independent rnod- ules. One of published applications of AIDA is dedicated to Conceptual Design of Aircraft. GPAL (Green Product All Life-Cycle) is a CAD system that integrates conceptual and detailed design. The kernel of the system consists in four modules: o functional element library, o a knowledge-based "Function to Form" mapping mecha- nism. o an assemblv model. Acta Polytechnica Vol. 43 No. 412009 Component providing Subfunction 1 Fig.5: Decomposirion of a global function into sub-funcrions o a module for geometric reconstitution of functional carri- ers based on default geometric reasoning [15]. 'I'he verifi- cation algorithms work with abstract features (conceptual geometric data structures are defined). All the above systems demonstrate formal facilities which are added to the "classical" loop ofconceptual design ("pro- posals-Evaluation-Corrections") and show still conrinuing differentiation of the abilities of such sysrems. B. "Prototyping" AI CDP Support Systems. The protoryping approach was frequently quoted and applied, especially in the late 1980s. The kernel of this ap- Fig. 6; Ontology - OMT - UML - CASE 8 proach consisted in an image of a functionally representative but uncompleted product of CDP which was sufficient for verification and for iterative corrections of the design. The greatest advantage of prorotype methods was in verifiing the specification requirements, which was independent of the completion of the detailed design stage. The prototyping approach was implemented very widely from CAUQ4114 systems with physical prorotypes of products to sophisticated program prototypes in the Software Engi- neering field. Fig. 6 illustrates an application where OMT - UIVIL - CASE are linked in CDP Working with OMT method- ology, the formulation of the problem environment and the P @@Ljryl@ryr Effiw =_- E€EE4@@#ffi Acta Polytechnica Vol. 43 No. 412003 problem solution goals are "translated" into the UML model. This model may be considered as a preliminary stage of the conceptual design process. If the rules for developing the for- mulation are detailed ar rhe necessary level (e.g., in rhe speci- fication of simulation, visualisation and measurement tools) and an appropriate CASE system with a good generaror of code is used for UML model, the results of CDP are repre- sented by means of the functions and the results of the generated programs. (-fhis way is surprisingly easily acces- sible, and it is not limited only to the conceprual design of systems which end at the level of programs.) C. Interpretati,on AI CDP Support Systems A traditional form of systems which support interpreta- tion is a set of rules. In a more detailed proposal ir is berter ro speak about a multi-view approach combined with a gradual knowledge acquisition procedure. One such system is MMfoTTED [24], which was developed to support the acquisition ofseveral ontologies for reasoning about an artefact from diflerent viewpoints. The system works with a hypertext browser which enables the designed objecr (situation, system) to be identified with pre-formed models. Browsing through the nenvork of models induces simulta- neous changes of views. The system integrates model-based reasoning, ontological engineering and hypermedia and web-based instruction. The system facilirares the exploration of design situations from different conceptual perspectives and supports problem setting and design development. D. Support for Emergent Phenornerm within CDP Though the emergence of novel solutions as outcomes of CDP is widely expected, the known means for formalisation and computer (or other) support for emergence phenomena in the field of CD have not been too effective. Promising projects in this field are usually covered by research support for creative phenomena. A relatively old approach (still quoted and still being developed) which may be included in the field of Conceptual Design on the left sid,e isformal and contputer suppod for TRIZ nuthodologl [l0]. This method works with a set of rules which have a formal representation of a technical problem (one of the formal tools for describing the problem is the "Sub- stance-Field" language and calculus) and on the right side a description of the solution operations. The operations are de- scribed as general heuristics, and the solution is achieved by interpreting these heuristics in the conceptual environment of the solved problem. This interpretation procedure repre- sents a special semiotic process which may be associated with an emergence phenornmon. This approach and method may serye as an experimental space for investigating emergence phenomena, though the set of rules (acquired by analysing an enormous number of creative technical solutions) is rather large (more than 1200 in [0]). A fonru,l descnption of emergmce cond;itions within CDP was introduced in [3]. The authors of this work described the emergence phenomenon by a context-preserving morphism from an original semiotic algebra to another semiotic algebra and by a condition claiming that the discovered solution is not in the co-domain of actions of the interpretation of the original algebra. This contribution focuses on explaining the emergent results acquired in an intuitive creative way, and the user application of rhis approach it is only illustrative. Another formal approach leading to a computer supporr toolfor thz emzrgmtplunomern within a co-eaotutiomry aarinnt of CDP has been presented in [28, 29]. Finally, a very promising line of research inro emergent phenomena within CDP is the multi-agrnt approach, where the emergence of a novel conceptual sglution is achieved as a result of information interaction of agents. The field of Conceptual Design was included as an experi- mental platform in the project "Methodology of Emergent Synthesis", as one of four parts of a project on "science of Synthesis" (Hi. Yoshikawa,Japan, 1996). The essenrial ideas of Enrcrgent Synthesis are explained, e.g., in [35]. The general direction of Ernergent Synthesis is slightly different fiom the approaches to the processing of emergence phenomena introduced above. 7 Conclusions The answer to the question "To whom belongs Concep- tual Design?" is that present-day research is concentrated on noael pdnciples of designing (relared to rnodelling of the mind), on ontologies (for semantical modelling), on intelligent took for computer support of CDP and on ernergent solution theory. Acknowledgements This research has been conducted at the Instirute of Instrumentation and Control Engineering, FME, CTU in Prague and has been supported by Research Grant GACR 102t01t0763. References tll Deng, Y. M., Britton, G. A., Tor, S. B: Constraint-based Functianal Design Verifuation for Canceptrnl Design. Com- puter-Aided Design, Vol. 32, 2000, p. 889-899. [2] Jansson, J., Horvath, l.: Beh,aaioral Sinrukttiun of In- cornplete Representations in Conceptual Design. In: Proc. of the l5tn European Simulation Multiconference (Ed.: Gerril, P.), Delft: SMS Press, 2001, p.321-325. t3l Roy, U., Pramanik, N., Sudarsan, R., Sriram, R. D., Ly- ons, K. W.: Function-to-fonn Mapping: Model, Representa- tion and Application in Design Synthesis. Computer Aided Design, Vol. 33, 2001, p. 699-719. t4l Taura, T.: Design Science for Functional Design Process Modelling. In: Proc. of l0th Int. Conf. on Engineering Design - ICED 95. (Ed.: Hubka, V.) Edition Heurista 95, Vol. 2, 1995, p.456-464. Suh, N. P.: The Princzples of Datgn. Oxford Series on Advanced Manufacturing. New York: Oxford University Press, 1990. O'Sullivan, B: Constraint-Aifud Conceptual Design. Lon- don (UK): Engineering Research Series, 2002. Tay, F. E. H., Gu, J.: Product Modelling for Conceptual Design Support Computers in Industry, No. 48, 2002, p. 143-155. Rentema, D., Jansen, E.: An AI Tool for Conceptual Design of Complex Produrts.ln: Design Research in Netherlands t5l t6l t7l t8l Acta Polytechnica Vol. 43 No. 412002 2000 (Eds.:Achten, H., Bauke de Vries andJim Flennes- sey). TU/e,2000, p. ll9-132. t9] Cottam, R, Ranson, W., Vounckx, R: Emergmce: Half of QnnturnJurnp ? Acea P olytechnica Scandinavica, Math- ematics, Computation and Management Engineering. Ser., 1998, p. 12-19. [0] Savransky, S. D.: Engineering of Creativitl: Intratluction to TRIZ Methodology of Inuentile Problem Sola'ing. New york: CRC Press, Boca Raton, 2000. Il] Opio, E. Z, How6th. I., Vergeest, J. S. M.: etalitl As- surance of Daign Support Softuare: Reaiew and. Ana$sis of the State of An. Computers in Industry, No. 49, 2002, p.195-215. [2] Kitamura, Y., Sano, T., Namba, K., Mizoguchi, R.: A Functiornl Concept Ontal.ogy an^d its Applir.afion to Auto- matic Idmtifuation of Functional Structures. Advanced En- gineering Informatics, No. 16, 2002, p. l4b-162. [3] Kryssanov, V. V., Tamaki, H., Kitamura, S.: Understa,nd- ing Design Fundamcntak: Hou Synthzsis and, Analjsis Driae Crealiaity, Resulting in Emngmce. Artificial Intelligence in Engineering, 2001, No. 15, p. 329-342. [4] Panowski, E.: Meaning in the Visml z4rts. Garden City, New York: Doubleday & Company, Inc., 1955. [5] Xu, Z. G., Tang, M. X., Frazer, J. H: From Concept to Embodimmt: Challcnge anl. Strategl. In: Proc. of Int. Conf. Artificial Intelligence in Design '02. (Ed.: J. Gero), Dordrecht: Kluwer Academic Publisher, 2002, p.215-236. [6] Rumbaugh,J.E, Blaha, M., Premerlani, W.J., Eddy, F., Lorensen, W.: Ohject-Uiznted, Modclling and, Des,ign. New York: Prentice-Hall Internarional Inc., 1991. [17] Rumbaugh, J. E., Jacobson, L, Booch, G: The Unzfud Mofulling Longnge Refermce Manttnl. Reading, MA: Ad- dison Wesley, 1999. [8] Sharpe,J. E., Bracewell, R., H.: Thz Llse of Funttional Rea- soning for thz Conrephnl,Design of Intndisciplirnry Schcmzs. In: Proc. of l0tn Int.'Conf. on Engineering Design - ICED'95. (Ed.: Hubka, V.), Edition Heurista 95, Vol. 2, 1995, p. 465-470. [19] Olesen, l.: Strengthnting tlw Undzrstnnting of Concephnl Duign.In: Proc. of l0'" Int. Conf. on Engineering De- sign - ICED'95. (Ed.: Hubka, V.), Edition Heurisra 95, Vol. 2, 1995, p. 47 l-476. [20] Bfla,J., Weiglhofer, W.: Multimodclling of La,rge Scale Sys- tems in Pre-Implementation Stage. ln: Proc. of Int. Conf. on Modelling and Simulation 2002. Universidad de Girona Press, 2002, p. 303-310. [21] Peckham, J., MacKellar, B.: Generating Cofu for Engi- neering Destgn Using Sof.ware Patterns. Artificial Intelli- gence in Engineering, No. 15,2001, p. 219-226. [22] Ikishnan, P.; Consistmcy Chcchs for UML. In: Proc. of 7tn Asia-Pacific Software Engineering Conference, IEEE, 2000, p, 162-169. [23] Bila, J.: The Optinaliry of UML Modzl Structure? Acta Mechanica Slovaca, No. 2, 2002, p.37-44. [24] Toppano, E.: MMfoTTED: A Cognitiue Tool Fostering thz Acquisitirn of Canuptunl Knowl.edge about Design products. In: AI in Design '02. (Ed.: J. Gero). Dordrecht: Kluwer Academic Publisher, 2002, p. l9l-214. [25] Pfanzagl , l.: Theory of Measurement \4rtrzburg-Wien: Physica-Verlag, l97l . [26] Ticlry, P: The Fourulntions of Frege's Logic. de Gruyter, 1988. [27] Jelinek, I.: Mofu.Iling of Design Process with Modat l-ogir.In: Proc. of the l5tn European Sirnulation Multiconference (Ed.: Gerril, P.), Delft: SMS Press,200l, p. 361-965. [28] Bila, J.: Emergent Phenomzna in a Co-Eaolutimary Concep- hutl Design. In: AI in Design '02. (Ed.r J. Gero). Dordrecht: Kluwer Academic Publisher, 2002, CD-ROM, paper No. l. [29] Bila, J.: Degree of Ernergence in Eaolutionary Sythcsis pro- cess. In: Int. Conference on Neural Networks - NN '2000, (Ed.: M. H. Hamza), Calgary: IASTED Press, 2000, p. 142-147. [30] Brasethvik, T., Gull, J.: Natural Innguage Analysisfor Se- ntantir Docunrcnt Modclling. Data and Knowledge Engi- neering, No.38,2001, p. 45-62. [31] Guarino, N., Giaretta, P.: Ontologizs and Knowledge Ba.ses, Towards a Terminologb.al Clnrifration. 1995, http//www.ladseb.pd.cnr. iVinfor/Ontolo gy/Papers /KBKS95.pdf. [32] Dartigues, CH., Parisa, G.: Product Datn Exchnnge Llsing Ontologics. In : AI in Design'02. (Ed. : J. Gero). Dordrechr: Kluwer Academic Publisher, 2002, p. 617-637. [33] Sloman, A: Inturacting Trajectories in Design Space and, Nichc Space: A Philosopher SpecuLa,tes About Eaolution. In: Proc. of 6tn Int. Conf. on Parallel Problem Solving from Nature. September 18-20, Paris (France), 2000, p. 19G-207. [34] Gruber, T.R.: Ontolingn: A Mech^onismto Support Pofiable Ontologi.es. Stanford University, Knowledge Systems Lab- oratory, 1992. [35] Ueda, K.: Emcrgent Synthesis. Arrificial Intelligence in En- gineering, 2001, No. 15, p. 319-327. [36] Materna, P.: The World, of Concepts and. Logic. Philosophy, Prague (Czech Republic), 2000. [37] Th6, E.: The Magb of M. C. Escher.I-ondon: Joost Elflers Books, Thames & Hudson, Ltd., 2000. Prof. Ing.Jiii Bfla, DrSc. phone: +420 224 352 534 fax: *420 233 336 414 e-mail: bila@fsid.cvut.cz Czech Technical University in Prague Faculty of Mechanical Engineering Technick6 4 166 07 Prague 6, Czech Republic t0 Scan1 3 Scan1 4 Scan1 5 Scan1 6 Scan1 7 Scan1 8 Scan1 9 Scan1 10