An Ontology-based Knowledge Management System for Industry Clusters

An Ontology-based Knowledge Management System for Industry Clusters Pradorn Su reephong 1 , Nopasit Chakpitak 1 , Yacine Ouzrout 2 , Abdelaziz Bouras 2 1 Departm ent of Knowledge M anagement, C ollege of Art s, Media and Techn ology, Chiang Mai University, Chian g Mai, Thailand. {dorn | nopasit}@camt.info 2 LIESP, University Lumiere Lyon 2, Lyon, Fra nce, {yacine.ouz rout | abdelaziz.bouras}@univ- lyon2.fr Abstract Knowledge-ba sed econom y forces com panies in eve ry country to group t ogether as a cluster in order to maintain their competitiveness in the world market. The cluster development relies on tw o key success fac tors whic h are knowled ge sharing an d collaboration between the actors in the clus ter. Thus, our study tries to propose a knowledge management syste m to support knowledg e management activities within the cluster. To achi eve the objectives of the st udy, onto logy takes a very important role in t he knowl edge ma nagement pr ocess in va rious wa ys; such as building r eusable and f aster knowledge-b ases and better ways of r epresenting th e knowledge explicitly. Howev er, crea ting and represen ting ontology creates difficulties to organization due to the ambiguity and unstructured nature of the source of knowledg e. Therefore, the objectives of this paper are to propose the methodolo gy to capture, create and re present ontol ogy for organization development by using the k nowledge e ngineering ap proach. The ha ndicraft cl uster in Thailand is used as a case study to illustrate o ur proposed methodology. Keywords: Ontolo gy, Semantic , Knowledge Manage ment System, Ind ustry Cluster 1.1 Introduction In the past, the three produc tion factors (Land, Labor and Capital) were abundant, accessible and were considered as the re as on of econom ic advantage, knowledge did not get m uch attent ion [1] . Nowadays, it is the knowle dge-based ec onomy era which is affect ed by the incre asing use of in formation t echnologies . Thus, previ ous production factor s are currently no long er enough to su stain a firm’s competitive advantage; k nowledge is bein g called on t o play a key role [2]. Most ind ustries try to use available information to g ain more competitive advantages than others. Knowledge-ba sed econom y is based on the prod uction, distri bution an d use of knowledge and in formation [3]. The study of Yoong and Molina [1] assu med that one way of sur viving in to day’s turbule nt business en vironment for business organizations is to form strategic alliances or mergers with oth er similar or 2 P. Sureephong, N. Chakpitak, Y.Ouzrout and A. Bouras complement ary business com panies. The conclusio n of Yoong a nd Molina’s study supports the idea of industry cluster [3 ] which is proposed by Porter in 1990. The objectives of th e grouping of firms as a cluster are maintainin g the collaborati on and sha ring of kno wledge am ong the part ners in orde r to gain competitiveness in their market. Therefore, Knowledge Management (KM) becomes a critical activity in achiev ing the goals. In order to manag e the knowledge, ont ology plays an importa nt role in ena bling the p rocessing and sharing of k nowledge bet ween experts and knowledge users . Besides, i t also provides a share d and comm on unders tanding of a domai n that can be communicat ed across people and applicat ion systems. On the other ha nd, creating ontology for an industry clu ster can create difficulties to th e Knowledge Engineer (KE) as well, because of the complexity of t he structure and tim e consumed. In this paper, we will propose the methodology for ontology creation by using knowledge engineering m ethodology in the in dustry clust er context. 1.2 Literature Review 1.2.1 Industry Cluster and Knowledge Mangement The concept of the industry cluster was po pularized by Prof. Mi chael E. Porter in his book “Competitive Advan tages of Nations” [3] in 1990. Th en, industry cluster becomes the current trend in economi c developm ent planning. Ho wever, there is considerable debate regarding the definition of the industry cluster. Based on Porter’s defi nition of i ndustry cl uster [4], the cluster can be se en as a “ geographica lly proximate group of companies and asso ciated institutions (for example universities, go vernment agen cies, and related asso ciations) in a particula r field, linked by co mmonalities and complementar ities ”. The general view of industry cluster map is shown in figure 1.1. Until now, literature of the industry cluster and cluster building has be en rapidly growing both in academ ic and policy-making cir cles [5]. Figure 1.1. Inustry Cluster Map Government A gents Cluster’s Co re Business Academic In stitutes Support ing Industries Associations CDA Ontology-based Knowledge Management System for Industry Cluster 3 After the concept of indust ry cluster [3] was tangibly applied in m any countries, companies in the same industry t ended to link to each othe r to maintain their competitiveness in their market and to gain benefits from being a me mber of the cluster. From the study of ECOTEC i n 2005[6] rega rding the cri tical success factors in cluster developm ent, the two critical success fac tors are collaboration in networkin g partnershi p and know ledge creation for innovative technol ogy in the cluster which are about 78% and 74% of articles mentioned as success criteria accordingly. This knowledg e is created through vari ous forms of loca l inter- organizational collaborative interaction [7 ]. They are collected in the form of tacit and explicit knowledge in experts and in stitutions within cluster. We applied knowledge en gineering tech niques to the ind ustry cluster in order to ca pture and represent the tacit knowled ge in the explicit form. 1.2.2 Knowledge Engi neering Techniques Initially knowledge engineerin g was just a field of the artificial intelligence. It was used to develop knowled ge-based systems. Until now, knowledg e engineers have developed their principles to improve the process of knowledg e acquisition since last decade [8]. These princi ples are use d to apply knowledge engineering in m any actual environment issues. Firstly, there ar e different types of knowle dge. This was defined as “know what” and “know how” [9 ] or “explicit” and “tacit” knowledge from Nonaka’s definition [10] Secondly, there are differen t type of experts and expertise. Third ly, there are many ways to present know ledge and use of knowledge. Finally, the use of structured method to relate the difference together to perform knowledge oriente d acti vity [11]. In our stu dy, many knowledge e ngineering methods have bee n compared [1 2] in order to sel ect a suitable method to be applied t o solve t he problem of industry cluster devel opment; i .e. SPEDE, MOKA, C ommonKAD S. We adopted CommonKADS methodology because it provides sufficient tools; such as a model suite (figure 1.2) and te mplates for di fferent knowledge in ten sive tasks. Figure 1.2. CommonKADS Model Suite Organization Model Task Model Agent Model Knowledge Model Communication Model Design Model Context Concept Artifact 4 P. Sureephong, N. Chakpitak, Y.Ouzrout and A. Bouras 1.2.2 Ont ology and Knowledge Managem ent The definition of ontology by Grub er (1993) [13] is “ explicit sp ecifications of a shared conceptualiza tion” . A conceptualiz ation is an abstract model of facts in the world by i dentify ing the rele vant co ncepts of the phenome non. Explicit means that the type of concep ts used and the constraints on their u se are explicitly defined. Shared reflects the notion that an ont ology cap tures consensual knowledge, that is, it is not private to the indivi dual, but accepted by the group. Basically, the role of on tology in the knowledge manag ement process is to facilitate the construction of a domain model. It provides a vocabulary of terms and relations in a sp ecific domain. In build ing a knowledge management system, we need two types of knowle dge [14]: Domain k nowledge : Knowledge about th e objective realities in the domain of interest (Objects, relations, events, states, ca usal relations, etc. that are obtained in some dom ains) Problem-solv ing knowl edge: Kn owledge abo ut how to use the dom ain knowledge to achieve various goals. This knowledge is often in th e form of a problem-solving method (PSM) that can help achieve th e goals in a different domain. In this study, we focus on ontolo gy creation a nd repres entation by ad opting knowledge engin eering methodo logy to suppor t both dimensions of know ledge. We use the ontology as a m ain mech anism to represent information and knowledge, and t o define the m eaning of term s used in the content language and the relation in the knowledge management system. 1.3 Methodology Our propose d method ology di vides ont ology int o three ty pes: generic ontology , domain ont ology and t ask ont ology. Generic ontology is the ontology w hich is re- useable across the doma in, e.g. organiza tion, product specification, c ontact, etc . Domain ont ology is the ontology defined for conceptualizing on th e particular domain, e.g. handicraft b usiness, logi stic, im port/export , marketi ng, etc. Task ontology is the ontology that specifies termi nology associated with the type of tasks and describes the problem solving st ructure of all the existing tasks, e.g. paper producti on, product shipping , product selecti on, etc. In our appr oach to im plement ontology- based knowl edge ma nagement, we integrated e xisti ng knowledge engineerin g method ologies and ontology development processes. We ado pted Comm onKADS for kn owledge engi neering methodol ogy and OnTo Knowled ge (OTK) m ethodology for ontol ogy development . Figure 1.3 shows t he assimilation o f Comm onKADS and On-To- Knowledge (OTK) [15]. Ontology-based Knowledge Management System for Industry Cluster 5 Figure 1.3. Steps of OTK methodology and CommonKADS model suite 3.1 Feasibility Study Phase The feasibility study serves as decision support for an economical, technical and project feasibility stud y, in order to select the most promising focus area an d target solution. This phase identifies p roblems, opportunities and potential so lutions for the organizati on and envi ronment . Most of the knowle dge engineerin g methodolo gies provi de the analysis m ethod to analy ze the organizati on before t he knowledge engi neering proc ess. This helps the knowledg e engineer to unde rstand the environm ent of the orga nization. C ommonKA DS also provides con text levels in the model suite (figure 1.2) in order to analyze organization al environment and the corresponding critical success factors for a knowledge system [16]. The organization model provides fiv e worksheets for analyzin g feasibility in the organizati on as sh own in fi gure 1.4. Figure 1.4. Organization Model Worksheets The Knowledge engineer can utilize OM-1 to OM-5 worksheets for interviewi ng with kno wledge decisi on makers of organizat ions. Then, the output s OM-1 Worksheet Problems, Solutions, Context OM-2 Worksheet Description of organization focus area OM-3 Worksheet Process breakdown OM-4 Worksheet Knowledge assets OM-5 Worksheet Judge Feasibility Feasibility Feasibility Study Ontology Kick Off Refinement Evaluation Organization Model Task Model Agent Model Knowledge Model Communication Model Maintenance and Evolution Design Model Feedback 6 P. Sureephong, N. Chakpitak, Y.Ouzrout and A. Bouras from OM are a list of knowledge intensiv e tasks and agents wh ich are related to each task. Then, KE c ould interview experts in each task using TM and AM worksheets for the next step. Finally, KE va lidates the result of each m odule with knowledge deci sion makers again to a ssess im pact and changes with the OTA worksheet. 1.3.2 Ont ology Kick O ff Phase The objective of this phase is to model the requ irements specification for th e knowledge m anagement system i n the organi zation. The Ontology R equirem ent Support Doc ument (OR SD) [17]gui des knowled ge engineers i n deciding abo ut inclusion and exclusion of c oncepts/relati o ns and the hierarch ical str uctur e of the ontology. It contains useful info rmation, i .e. Domai n and goal of the ont ology, Design guidelines, K nowledge sour ce, User and usage scenario, Com petency questions , and Appl ication supp ort by t he ontology [15]. Task and Agent Model are separat ed in to TM-1 , TM-2 and AM worksheets. They facilitate KE to complete the ORSD. The TM-1 worksheet identifies the features of re levant tas ks and knowle dge sourc es availabl e. TM-2 w orksheet concentrates in detail on bottleneck and improvement relating to specific areas of knowledge. AM worksheet lists all relevan t agents who possess knowledge items such as dom ain expert s or knowl edge w orkers. 1.3.3 Refinem ent Phase The goal of the refinem ent phase is to prod uce a mature and appli cation-orie nted target ontology according t o the specificatio n give n by the kick off phase [18]. The main tasks in this phase are knowledge elicitatio n and formalization. Knowledge elicitation pro cess with the domain expert based on the initial input from the kick off phase is pe rforme d. Comm onKADS provi des a set of knowle dge template s [11] in orde r to supp ort KE to capt ure knowle dge in differe nt types of tasks. CommonKADS classify knowledge in tensive tasks in two categories; i.e. analytic tasks and synthetic tasks. The fi rst is a task regardi ng systems that pre- exist. In opposition, the synthetic task is about th e system that does not yet exist. Thus, KE should real ize about the type of t ask that he is dealing with. Fi gure 1.5 shows the different knowledge task types. Figure 1.5. Knowledge-intensive task types based on th e type of problem Knowledge Intensive Task Anal y tic Task S y nthetic Task Classificatio Dia g nosis Assessment Monitorin g Prediction Desi g n Plannin g Modelin g Schedulin g Configuration Desi g n Assi g nment Ontology-based Knowledge Management System for Industry Cluster 7 Knowledge formaliza tion is transformation of knowledge in to formal representati on languages suc h as Ontology Inference Laye r (OIL) [1 9], depends on application. Therefore, the knowledge engin eer has to c onsider the a dvantages a nd limitations of the different languages to select the appropriate one. 1.3.4 Eval uation Phase The main objectives of this phase are to ch eck, whether the target ontology suffices the ontology requirements and whether the ontology based knowledge management system support s or answers t he compet ency questions, a nalyzed in the feasibility and kick off phase of the project. Thus, the ontology shou ld be tested in the target applicati on envir onment. A p rototype sh ould alrea dy show core functionalities of the target system. Feed backs from users of the proto type are valuable in put for further re finem ent of the ontology . [18] 1.3.5 Mainten ance and Evoluti on Phase The maintenan ce and evolut ion of an ontol ogy-base d applicati on is primari ly an organizati onal process [1 8]. The knowle dge engineers have to up date and m aintain the knowledge and onto logy in their responsibility. In ord er to maintain the knowledge m anagement sy stem, an ontology editor m odule is de veloped to help knowledge engin eers. 1.4 Case Study The initial investigations have been done with 10 firms within the two biggest handicraft ass ociations in T hailand and N orthern Thail and. No rthern H andicraft M anufacturer and EX porter (NOHMEX) associatio n is the bi ggest handi craft association in Thailand wh ich includes 161 manufacturers and exp orters. Anot her association which is the biggest handicraft association in Chiang Mai is named Chiang Mai Brand which inclu des 99 enterprises. It is a group of q ualified manufacturers who have capability to exp ort their products and pass the standard of Thailand’s ministry of comm erce. The objective of this stud y is to create a Knowledge Management System (KMS) for supporting this handicraft cluster. One of the critical tasks to im plement this system is creating ontologies of the knowledge tasks. Because, ontology is recognized as an appropriate methodol ogy to accomplish a comm on consensus of communication, as well as to support a diversity of activities o f KM, such as knowledge re pository, retri eval, shari ng, and dissem ination [20]. I n this case, knowledge e ngineerin g method ology was ap plied fo r ontolo gy creation i n the domain of T hailand’s handicr aft cluster . Domain Ontology: can be created by using t hree models i n context level of model suite; i.e. orga nization model, task model and agent m odel. At the beginni ng of doma in ontology creat ion, we ado pt generi c ontology pl us acquire d inform ation from the worksheets as an outline. Then, the more information that can be acquired 8 P. Sureephong, N. Chakpitak, Y.Ouzrout and A. Bouras from organization and environment, the mo re domain-oriented ontology can be filled-in. Task Ontology: specifies terminology associated with the type of tasks and describes the problem solving structure. The objective of k nowledge engineering methods is to solve proble ms in a specific dom ain. Thus, m ost of knowledge engineering a pproaches pr ovide a collect ion of pre defined sets of model elements for KE [16] . Comm onKADS m ethodology a lso provi des a set of tem plates i n order to support KE to capture know ledge in differen t types of tasks. As shown in figure 1.5, there are var ious types of know ledge tasks that need different ontology. Thu s, KE has to select the appropriate template in order to capture right knowledge and ontology. For illustration, we will use classifica tion template for analytic task as an example for task ontology creation. Fi gure 1.6 shows the inferences structure for classification m ethod (left side) and task ontology (ri ght side). Figure 1.6. CommonKADS classifi cation te mplate and task ont ology In the case study of a handicraft cluster, o ne of the kn owledge inte nsive tasks is about product selectio n for exporting. Not all h andicraft products are ex portable due to their specifications, function, a ttributes, etc. Moreover, t here are many criteria to select a product to be exported to specific co untries. So we defined the task ontology of the product selecti on task (see the right side of figure 1.6). 1.5 Conclusion The most im portant rol e of ont ology in k nowledge m anagement is to enable an d to enhance knowl edge shari ng and re using. Moreo ver, it provi des a comm on mode of communicat ion among t he agents and knowledge engineer [1 4]. However , the difficulties of ontology creation are claimed in most literature. Thus, this study focuses on cr eating ontolog y by adopting the knowledg e engineering meth odology which provi des tools t o support us for st ructurin g knowledge . Thus, ont ology was applied to help Knowledge Management System (KMS) for the industry cluster to achieve their goals. The architecture of th is system consists of three parts, Ob j ect Class Attribu te Feature Truth Value Ge n e r at Specify Obtain Match Object Candidate Handicraft Product Export Product Non Export Product Feature Attribu te Ontology-based Knowledge Management System for Industry Cluster 9 knowledge sys tem, ont ology, and knowled ge engineering . Hence, the proposed methodolo gy was used to create ont ology in the handi craft clust er context. Duri ng the manipulation stage, whe n users accesse s the knowledge base , the ontology can support tasks of KM as well as searching. The knowledge base and the onto logy is linked one to anot her via the ont ology module. In the m aintenance stage, knowledge enginee rs or domai n experts can add, up date, revise, a nd delete the knowledge or domain ontology via knowledge acquisition module [21]. To test and validat e our approach an d architecture, we used the handicraft cluster in Thailand as a case study. In our perspectiv es of this study, we will finalize the specification of the shar eable knowledge/information and the conditions of sharing among the cluster me mbers. Then, we will capture and maintain the knowledge ( for reus ing knowledg e when requ ired) and work on the specific infrastructure to enhance the collaboration. At th e end of the study, we will develop the kno wledge managem ent syst em for the handic raft cluster rela ting to acquiring requi rements speci fi cation from the cluster. 1.6 References [1] Young P, Molina M, (2003) Knowledge Sh aring and Business Clusters, In: 7 th Pacific Asia Conference on Inform ation S ystems, pp.1224-1233. [2] Romer P, (1986) Increasing Return and Long-run Growth, Journal of Politi cal Economy, vol. 94, no.5, pp.1002-1037. [3] Porter M E, (1990) Competitive Advant age of Nations, New York: Free Press. [4] Porter M E, (1998) On Competition, Boston: Har vard Business School Press. [5] Malmberg A, Power D, (2004) (How) do (firms in) cluster create knowledge?, in DRUID Summer Conference 2003 on creating, sharing and transferring knowledge, Copenhagen, June 12-14. [6] DTI, (2005) A Practical Guide to Cluster Development, Repo rt to Department of Trade and Industry and the English RDAs b y Ecotec Research & C onsulting. [7] Malmberg A, Power D, On the role of gl obal demand in local innovation processes: Rethinking Regional Innovation and Change, Shapiro P, and Fushs G, Dordrecht, Kluwer Academic Publish ers. [8] Chua A, (2004) Knowledge ma nagement system architectu re: a bridge between KM consultants and technologist , International Journal of Informatio n Management, vol. 24, pp. 87-98. [9] Lodbecke C, Van Fenema P, Powell P, Co -opetition and Knowledge Transfer, The DATA BASE for Advances in Informa tion System, vol.30, no . 2, pp.14-25. [10] Nonaka I, Takeuchi H, (1995) The Knowle dge-Creating Compan y, Oxford University Press, New York. [11] Shadbolt N, Milton N, (1999) From knowledge engineering to knowledge management, British Journal of Manage1m en t, vol. 10, no . 4, pp. 309-322, Dec. [12] Sureephong P, Chakpitak N, Ouzrout Y, Neubert G, Bouras A, (2006) Economic based Knowledge Management System for SM Es Cluster: case study of handicraft cluster in Thailand. SKIMA Int. Conference, pp.10-15. [13] Gruber TR, (1991) The Role of Common Ontology in Achieving Sharable, Reu sable Knowledge Bases, In J. A. Allen, R. Fikes, & E. Sandewall (Eds.), Principles of Knowledge Representation and Reasoning: Pr oceedings of the Second International Conference, Cambridge, MA, pp. 601-602. 10 P. Sureephong, N. Chakpitak, Y.Ouzrout and A. Bouras [14] Chandrasekaran B, Josephson, JR, Richard BV, (1998) Ontology of Tasks and Methods, In Workshop on Knowledge Acqui sition, Modeling and Management (KAW'98), Canada. [15] Sure Y, Studer R, (2001) On-To-Knowledge Methodology, evaluated and employed version. On-To-Knowledge deliverable D-16, Institute AIFB, Un iversity of Karlsruhe. [16] Schreiber A Th, Akkermans H, Anjewerden A, de Hoog R, Shadbolt N, van de Velde W, Wielinga B, (1999) Knowledge Engineering and Management: The CommonKADS Methodology, The MIT Press. [17] Sure Y, Studer R, (2001) On-To-Knowle dge Methodology, final version . On-To- Knowledge deliverable D-18, Institute AIFB, University of Karlsruhe. [18] Staab S, Schnurr HP, Studer R, Sure Y, (2001) Knowledge processes and ontologies, IEEE Intelligent Systems, 16(1):26-35. [19] Fensel, Harmelen Horrocks (OIL) [20] Gruber T R, (1997) Toward principles for the design o f ontologies used for knowledge sharing, Int. J Hum Comput Stud, vol. 43, no. 5-6, pp.907-28. [21] Chau K W, (2007) An ontology-based know ledge management system for flow and water quality modeling, Advance in En g ineering Software, vol. 38, pp. 172-181.