A Description Logic for Composite Objects for Domain Modeling in an Agent-Oriented Application

In this paper we present a description logic for composite objects and show its use for domain modeling in a particular application. The description logic extends standard description logics by providing support for representation of and reasoning about part-of. We have extended the standard reasoning tasks of subsumption, classification and recognition to include knowledge about part-of. Further, we maintain a part-of hierarchy for individuals that can be used for querying the system about composite objects. We provide the system with new user functions to allow for traversing the part-of hierarchy as well. Our extended description logic system retains the well-known advantages of description logic systems while extending their representational and inferential capabilities. The application involves automating the monitoring of the Reaction Control System (RCS) of NASA’s space shuttle. The application has been modeled before and an implementation exists using the agent-oriented system dMARS (distributed Multi Agent Reasoning System) from the Australian Artificial Intelligence Institute. Although it was recognized that part-of played an important role in the application, the existing implementation did not provide any support in representing and inferencing with part-of. The use of our description logic for composite objects has provided a number of advantages. In the following section we briefly describe dMARS and the Reaction Control System application (section 2). We propose our description logic for composite objects and describe the advantages that the use of this description logic gave us for the RCS application in section 3. The paper concludes in section 4.