![]() |
|||
![]()
|
![]() |
![]() |
![]() |
5. ONTOLOGICAL COMMITMENTSA pipe is a long, round, hollow object, usually made of metal or plastic, through which a liquid or gas flows. But, in a smoker context, a pipe is an object used for smoking tobacco. When two plumbers who are also smokers talk and say "give me the pipe," they need to commit to which context they talking and, assuming a plumbing context, they need to know if they are talking about metal or plastic pipes. The same occurs when the plumber attends a course about new kinds of pipes; instructors and plumbers need to commit most of vocabulary and most of the meanings of those terms. Certainly, when intelligent systems reuse or share knowledge of others, they need also to know and to commit to such terms and their meanings; these terminological agreements are called ontological commitments. Using Gruber and Olsen's definition (Gruber and Olsen, 1994), "Ontological commitments are agreement to use the shared vocabulary in a coherent and consistent manner." They guarantee consistency, but not completeness. Ontological agreements are useful not only in distributed systems that share vocabularies, but also inside each system (Mark, Dukes-Schlossberg, and Kerber, 1995). In other words, any system needs to be coherent and consistent with itself throughout its entire life cycle. So, any knowledge added or modified in the KB needs to be coherent with the previous agreements established during its development. To follow on with the plumber example, if a new pipe is introduced into a KB, it must fit the ontological agreements about pipes. Is sum, we can say that any source of knowledge (KB or ontology) codified in a formal language has, explicitly or implicitly, its own commitments. Any extension to the ontology must satisfy these ontological commitments. 6. TYPES OF ONTOLOGIESThis section does not seek to give an exhaustive typology of ontologies as presented in (Mizoguchi, Vanwelkenhuysen, and Ikeda, 1995). But it is interesting to show the most commonly used types of ontologies so that the reader can get an idea of the knowledge to be included in each type of ontology. Basically, there are four kind of ontologies: domain ontologies, task ontologies, common sense ontologies, meta-ontologies, and knowledge representation ontologies. Domain ontologies provide a vocabulary for describing a given domain. They include terms related to:
Sometimes, if the domain ontology covers a large extension of objects, verbs, and primitive concepts, it is recommended to split the domain ontology into three categories of ontologies: object ontologies, activities ontologies, and field ontologies that cover, objects, activities, and primitive concepts respectively. Task ontologies provide a vocabulary for describing terms involved in problem-solving processes, which could be attached to similar tasks that may, or may not, be in the same domain. They include nouns, verbs, paraphrases, and adjectives related to the task. For example, the terms "goal," "schedule," "to assign," "to classify," "to plan," "assigned" and so on will be in a planning ontology. We recommend that nouns be bunched together in a generic nouns ontology, verbs in a generic verbs ontology, and adjectives in a generic adjectives ontology if the task ontology is wide-ranging. As mentioned above, ontologies are used to represent common sense knowledge such as time, space, causality, events, etc. These ontologies are called common sense ontologies and they include a wide-ranging amount of foundational knowledge. Prototypes of common sense ontologies are Cyc ontologies (Lenat and Guha, 1990). Meta-ontologies provide the basic terms used to codify either a domain ontology, a task ontology, or a common-sense ontology in a formal language. Finally, knowledge representation ontologies captures the representation primitives used in knowledge representation languages. For example, in Ontolingua (Gruber, 1993a), the knowledge representation ontology is the frame ontology. A meta-ontology could also provide the basic core of concepts upon which other ontologies are built.
|
![]() |
|
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. |