Design and Use of an Ontology for Complex Artifacts

During the past decade we have developed OCA, an ontology dedicated to complex artifacts. OCA has been applied extensively within the domain of communication systems and has proven a valuable tool for building models of both products and processes. Today, OCA is tightly integrated with a modeling method called MacroModelingTechnique (MMT). OCA-based MMT has increased the effectivness of model building particularly during the early phases of an industrial engineering process where numerous technical and non-technical aspects need to be considered and shared by diverse experts. This paper discusses and assesses the design of OCA. Much of OCA’s design hinges on a technique to construct key concepts in a principled rather than in an ad-hoc manner. In fact, much of OCA’s design is determined by "ontological principles". Aspects of Ontology Design "Ontology design" as yet is not a crisp and proven technical term. So we will begin by investigating several aspects of ontology design. We assume that an ontology has properties which can be or should be subject to design. To explain such properties we use OCA as an example. OCA is an ontology tuned to the domain of complex artifacts such as industrial products and processes. Generally speaking, OCA is a collection of useful concepts to describe such things. Some of its concepts have been customized to build models of communication systems. When considered more closely, OCA is a complex artifact by itselfa system with an elaborate structure including many components. OCA also has numerous characteristics and exhibits many interrelations with its rich and diverse environment. In a first step we will look at OCA’s internal structure in particular at its composition. Here we will identify key components which are key concepts. In a next step we will clarify and justify the structure and construction of those key building blocks. Later on we will investigate in what way and to what extent the component concepts of OCA interact and depend on each other. We will find strong interaction in OCA, which is in part due to prominent "root concepts" such as the SYSTEM concept which consitutes the "root" of many generalizations and specializations. In fact, the construction of root concepts influences the construction of most other concepts in OCA and thus determines OCA’s internal structure to a large extent. As a consequence, we need to carefully consider concept design when designing OCA. One aspect of concept construction is concept composition. Here we will show how the structure of a concept, and in particular a root concept, can be derived from "ontological principles". So far we have identified ontology structure as one of the important aspects and objectives of ontology design. In this context we have pointed to interaction, composition, the role of root concepts and the significance of structuring principles. But equally important aspects of design are key characteristics which apply to an ontology in its entirety. OCA, for instance has a key structural characteristic namely modularity. It also has a key functional characteristic, "relevance". By drawing on our experience with OCA, we assume that modularity, relevance, and other key characteristics determine the quality of an ontology to a great extent and thus should be subject 181 From: AAAI Technical Report SS-97-06. Compilation copyright © 1997, AAAI (www.aaai.org). All rights reserved. to design. In OCA, the design of root concepts is an example of a design for modularity. The design of "basic concepts", on the other hand, is an example of a design for relevance. In OCA, a concept qualifies as being basic if it meets certain criteria like level of abstraction, sophistication, and frequency of use. The SYSTEM concept will serve as a good example of a basic concept. Concepts like SYSTEM gain both modularity and relevance from "ontological principles". As a consequence, such structuring principles play a prime role in the design of OCA and may play that role for other ontologies as well. Lastly, we will focus on the aspect that an ontology needs to be designed for and fit into an environment. OCA for instance, does play the role of a tool which relates to particular tasks. One of those tasks is to describe a product in a comprehensive, simple, concise and canonical manner, and suitable for many kinds of users and uses of such descriptions. Design, thus, is only one phase in OCA’s life cycle which also includes phases of application and adaption based on practical experience. In this course, we have found it practical to regard OCA as a system where "system" may be defined in the way we have done in form of the SYSTEM concept which will be explained later