Ontological Engineering for B2B E- armyunis/knowledge-managment/papers...Ontological Engineering for...

Click here to load reader

  • date post

    16-May-2018
  • Category

    Documents

  • view

    217
  • download

    1

Embed Size (px)

Transcript of Ontological Engineering for B2B E- armyunis/knowledge-managment/papers...Ontological Engineering for...

  • Ontological Engineering for B2B E-Commerce

    Leo Obrst Robert E. Wray Howard LiuThe MITRE Corp.

    7515 Colshire DriveMcLean, VA 22102

    [email protected]

    [email protected] P. 0. Box 22285Sacramento, CA [email protected]

    Abstract - In this paper we discuss the nature of our overall enterprise to create ontologies inthe product and service knowledge space for Business-to-Business (B2B) electronic commerce.We descr ibe one crucial problem: the mapping problem, i .e . , mapping among ontologies ,taxonomies, and classification systems, some of which are more semantically sound and coherentthan others. This problem we consider to be in need of a sustained research program if tenableso lu t ions a re to be found , s ince the lack of a so lu t ion wi l l p rec lude widespread adopt ion ofontologies by the commercial world. Final ly, we summarize the general issues we faced andindicate prospective future research.

    Categories 8 Descriptors - 1.2.4 [Artificial Intelligence]: Knowledge RepresentationFormalisms and Methods - Predicate logic, Representation languages, Representations(procedural and rule-based), Semantic networks.

    General Terms - Algorithms, Design, Economics, Experimentation, Languages,Management, Performance, Reliability, Standardization, Theory.

    Keywords -Ontological engineering, ontology management, ontology mapping, ontologymerging, business-to-business electronic commerce, B2B e-commerce.

    1. IntroductionIn this paper we discuss the nature of our overall enterprise as ontologists and domain expertsworking to c rea te onto logies in the product and serv ice knowledge space for B2B e lec t ron iccommerce that included domains ( lower ontology) , a gener ic upper ontology adapted from theCyc upper model, (http://www.cyc.com/tech.html) and shared middle ontologies . We descr ibeone crucial problem: mapping, i .e . , mapping among ontologies , taxonomies, and classif icat ionsystems, some of which are more semantically sound and coherent than others. We consider thisproblem to be in need of a sustained research program, since the lack of a solution will precludewidespread adoption of ontologies by the commercial world.

    The structure of this paper is the following. In Section 2 we describe the B2B enterprise and theuse of ontologies to facilitate this enterprise. In Section 3 we highlight the mapping problem.

    Permission to make digital or hard copies of all or part of this work for personal or classroom use isgranted without fee provided that copies are not made or distributed for profit or commercialadvantage and that copies bear this notice and the full citation on the first page. To copy otherwise, orrepublish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.FOIS'OZ, October 17-19,2001, Ogunquit, Maine, USA.Copyright 2001 ACM I-58113-377-4/01/0010...$5.00.

    117

  • Finally, in Section 4 we summarize the general issues we faced and indicate prospective futureresearch.

    2. The Nature of the B2B EnterpriseB2B electronic commerce is everything that land commerce is, plus more: automated support forinformat ion and t ransact ion f low and for ver t ical and hor izontal commercia l in teroperabi l i ty .B2B electronic commerce includes the following: multiple marketplace platforms on the Internetthat suppor t mul t ip le t rading models (auct ions , reverse auct ions , exchanges , Request-For-Proposal/Request-For-Quote (RFPRFQ), bookstores, trading hubs, etc.) for and by commercialorganizations, providing rich information content on products and services for both buyers andsel lers (ca ta logs , product guides , market and domain edi tor ia l content , news, adver t i s ing) andsupport for buying nd selling, financing, privacy/security, payment processing, ordermanagement, protiling/personalization, product configuration, planning/scheduling andforecast ing, product l i fe cycle and inventory management , business processes , workflow, andrules, logistics, distribution, and delivery.

    B2B e-commerce needs ontologies. First, there is an informational need: because the ontologyis a structured conceptual model of the e-commerce vert ical domain (and sometimes, quasi-hor izonta l domains too) , i t suppor ts parameter /proper ty-based search and naviga t ion us ingproduct and service knowledge by prospective buyers to discover what to buy, andsubsequently to determine pricing and availability. In this case, the relatively static knowledgeof the ontology maps to the relatively dynamic data of the vendors. Furthermore, an ontologycan model not only commodit ies , but a lso agents , i .e . , buyers and sel lers , both human andartificial. By employing user role knowledge (sometimes called user profiling orpersonal iza t ion) to ass i s t the search process , quer ies can be cus tomized to a user s knownfunctions and interests, possibly based on previous interaction with that user.

    E-commerce also needs ontologies for t r a n s a c t i o n a l purposes : knowledge of a companysorganiza t ional s t ruc ture , workf low, processes , and products / serv ices can be used to ac tua l lyassist in buying and selling directly.

    The ontologies were built to support the representation requirements, as opposed to thepresen ta t ion requirements of the intended applicat ions, al l of which presumed some form ofclassification of products and services. Classification systems are typically ad hoc, inconsistent,and not integrated, with little association between classification systems. Representation was theunderlying structure and codification of the product and service knowledge space to be suppliedby the even tua l ly deve loped on to log ies . This represen ta t ion would be semant ica l ly sound ,consistent (though obviously always incomplete because additional refinements could always bemade) , control led, modular , reusable , and provide some support for appl icat ion presentat ionneeds. Presentation was largely the responsibility of the application, though the ontologies wouldoffer some support . I t i s in the mapping among ontologies and c lass i f icat ion systems, and themapping among representa t ion categor ies and those presenta t ion categor ies preferred byparticular applications, that technical difficulties lie. This mapping problem we address in thenext section.

    3. Mapping Ontologies to Ontologies, Taxonomies and ApplicationsIn th i s sec t ion we d i scuss in de ta i l one c ruc ia l i s sue we faced in deve lop ing on to log ies tosuppor t B2B e-commerce: the problem of mapping reference ontologies with well-definedsemant ics to o ther ontologies , taxonomies , and s tandards-based c lass i f ica t ion sys tems that areless semantically sound and coherent.

    118

  • Ideal ly , ontologies provide a semant ic infras t ructure that can be used for a l l appl icat ions . Toprovide this semantic framework across the appl icat ions/data of an e-commerce business( independent ly developed and wi thout commitment to ontologica l inf ras t ruc ture) , ex is t ing andplanned information resources must be connected to the ontology framework. We term thisconnect ion a mapping. A mapping is a many- to-many re la t ionship between source data andan ontology. Sources for mapping could be another ontology, some standard taxonomy, OT anapplications data structures.

    . ..~. ....~ 1.. . ~.

    1 ,_, , ._ . , ,_.: . .. i- _. ., , .

    Figure 1. A Simple, Informal Application Taxonomy (left) Mapped to an Ontology(right)

    Figure I illustrates a mapping. Solid lines indicate a well-defined subclass relation in theontology (right); in the taxonomy (left) the solid line represents a parent-child relation, with ill-defined semant ics . The heavy, dotted lines without arrows represent other ontological relations.The thin, double-armwed l ines dep ic t a mapping be tween nodes in the on to logy and da tastructures in the application taxonomy.

    On the left, an e-commerce application uses a taxonomy with ill-defined semantics to representsome information. For example, node Z could represent some indus t r ia l p rocess . Node Y cou ldrepresent products resulting from this process, X the equipment used in the process, and W theemployees involved in the process . The re la t ion be tween nodes i s an undef ined parent -chi ldre la t ion wi th no inher i t ance ; the in format ion i s used in the app l ica t ion to group these re la tedconcepts together. On the right, an ontology represents much of the same information, but with awell-defined semantics for each relation. For example, nodes B and C could represent productsresulting fmm the industrial process A. The relation between A and these nodes is not subclass;it might be generated-&om, assuming the semantics of this relation was defined. B e c a u s e Band C are products, they are subclasses of a more general pmdnct node iJVQ, perhaps in a middleor upper ontology. Unlike the application taxonomy, a different, well-defined relation relates Ato D, and so on.

    One criticism of this scenario is that the application taxonomy is deficient. In other words, whatis needed is not a mapping, but a clearly defined semantics in the application taxonomy. W i t hsuch a def ini t ion, some automated merging process could (conceivably) merge the nodes.However, practically, such merging was not possible.

    Mappings pmvide an intermediate solution to this problem. Once Z is mapped to the ontology,other applications can recognize that Z is similar to A. This example is only one of severaldifferent kinds of