![]() |
|||
![]()
|
![]() |
![]() |
![]() |
15. SYSTEMS THAT USE ONTOLOGIESThis section briefly summarizes a few examples of how to use ontologies in real applications, like software systems design, engineering negotiation, unification of database schemata, etc. Comet and Cosmos were developed by Mark, Dukes-Schlossberg, and Kerber (1995). Comet supports the design of software systems, and Cosmos supports engineering negotiation. Both systems give design feedback to their users. When a Comet user modifies a software module, Comet provides feedback on which other modules are affected and will require modifications. Cosmos provides hardware designers with analyses that indicate the impact of a proposed design change. In both systems, when a new software module is created and when new design changes are supplied, the changes create new LOOM concepts that are added to their knowledge bases. In both systems, it is essential that the commitments to be satisfied by an additional be encoded in the original core of the KB in order for it to be consistent and for the system's reasoning methods to act properly. The goal of the TOVE (TOronto Virtual Enterprise) Enterprise Modeling project (Grüninger and Fox, 1995) is to create enterprise models that not only answer queries using what is explicitly represented, but is also able to deduce answers to queries. The goal of the Enterprise Project (Uschold and Grüninger, 1996), developed by Uschold at AIAI in Edinburgh, is to improve and where necessary replace existing modeling methods with a framework for integrating methods and tools appropriate to enterprise modeling and the management of change. The Enterprise ontology plays an important role in this project. This ontology includes: a meta-ontology and ontologies related to activities and processes, strategies, organizations, and marketing. KACTUS (Schreiber, Wielinga, and Jansweijer, 1995) is an ESPRIT project on modeling knowledge about complex technical systems for multiple use and the role of ontologies to support it. Ontologies in the domain of electrical networks, offshore oil production, and ship design and assessment have been already built. Plinius (Vet, Speel, and Mars, 1995) is a semi-automatic knowledge acquisition system from natural language text in the domain of ceramic materials, their properties, and their production processes. The Plinius ontology is one of the cores of the Plinius system, because there exist many other things that make the system work. The Plinius ontology provides the semantics of the terms used by the Plinius system lexicon. This ontology might be classified as a domain ontology. Top-down and bottom-up methods have been merged to build the ontology. One of the features of the Plinius ontology is that is stable and fixed, so that it can serve as an anchor point for the processes. Top-down methodologies are quite useful to provide the core of such concepts. In systems like Plinius, it is quite difficult to exhaustively predict in advance which concepts you are going to need. So, one of the main requirements was to be able to specify beforehand each and every possible substance. As the list of possible substances is infinitely large, this can only be done implicitly using bottom-up methods. The design criteria followed in developing the Plinius ontology are: parsimony, faithfulness (i.e., domain experts would have to recognize the ontology as a correct rendering of the way their domain is organized); easy extendibility; limitableness; and relevance of the formal definitions and informal explications. The Plinius ontology was conceptualized on the principle of a conceptual construction kit. According to this principle, an ontology consists of atomic concepts and a set of construction rules (also called transformation rules) that define other concepts. Nonatomic concepts are implicitly defined in the ontology. The ontology can be seen as a calculus. The deductive closure of this calculus produces the explicit list of concepts. In Plinius, atomic concepts are: chemical elements, which are defined by their atomic number, and natural numbers, defined in a number theory. The Plinius ontology has been implemented in a wide range of representation languages and systems, such as: Ontolingua, CLASSIC, and Prolog (the Prolog implementation is the most satisfactory). This work is a framework of reference about advantages and disadvantages of choosing representation languages for ontologies. According to Plinius conceptualization, not all the conceptualized knowledge is easily implementable in the chosen target language since each language has its own expressiveness. Since large parts of the ontology are bottom-up, it is difficult to codify it in target languages that assume top-down ontologies like Ontolingua. In this case, most of the definitions in Ontolingua had to be written in pure KIF. As a result, the ontology cannot be translated into target languages like LOOM and EPIKIT because the automatic translation only works for pure Ontolingua. More detailed information of this system, including the Ontolingua ontology, can be examined at the following URL: http://wwwis.cs.utwente.nl:8080/kbs/ontology/homepage.html. Dowell, Stephens, and Bonnell (1995) present an ontology of domain knowledge that can be used as a semantic gateway between different database schemata. The ontology allows the user to query and get information from nonlocal databases They first map each database schema to an extended conceptual model and link each part of the extended conceptual model with its corresponding items in the domain ontology. The domain ontology consists of two hierarchies: a hierarchy of types that correspond with entities in the database, and a hierarchy of links that represent relations and properties in the database. 16. CONCLUSIONThis chapter presented an overview of the knowledge sharing and reuse technology, as well as METHONTOLOGY, a set of guideliness that allow to build an ontology since an engineering point of view, reducing the existing gap between the ontological art and ontological engineering. The main future trends include:
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |