Augmenting Agent Knowledge Bases with OWL Ontologies

The use of software agents, viewed as conditioned sets of automatic functions, has become a common and important tool in modem software engineering. Intelligent software agents, are agents that contain some model of the environment that includes a representation of goals desirable to the agent, and that are able to reason about the actual situation and determine some appropriate course of action to realize the agent's goals. The model of the environment is also referred to as a knowledge base. We have been developing intelligent software agents to support requirements analysis for autonomous military UAV systems. The agents are developed to provide complex and competent opponents as well as representative prototype UAV's with varying degrees and types of autonomous functionality. Both types of agent are implemented using the AOS JACK Agent Framework. The agents used in JACK model reasoning behavior according to the Belief Desire Intention (BDI) model of artificial intelligence. JACK intelligent agents are autonomous software components that have explicit goals to achieve or events to handle (desires). To describe how they should achieve these desires, BDI agents arc programmed with a set of plans. Each plan describes how to achieve a goal under varying circumstances. Set to work, the agent pursues its given goals (desires), adopting the appropriate plans (intentions) according to its current set of data (beliefs) about the state of the world. In a previous report, we described the use of a suite of OWL ontologies as a basis for interoperation among a collection of legacy and prototype engineering analysis and design systems. In this paper, we discuss the application of that same set of ontologies as a knowledge base for intelligent software agents that operate in the environment created by the collected systems. We describe a "layered" set of OWL ontologies that primarily serve as a domain model for a number of legacy operations analysis simulations, aeronautical engineering design tools and software prototypes used to support the integrated design and development of aeronautical systems. We use these ontologies to provide a source for the derivation of agent-specific knowledge bases. (i.e. concepts, or in the case of JACK agents, "belief sets" are read directly from specified concepts and individuals in the ontology). The agents then use these facts to populate the agent's world model and subsequently, operate on these facts. The ontologies also provide a domain-model or context for the agent-specific knowledge base that allows access to facts about the "world" in which the agent is situated. Thus, concepts that are not directly part of the agent model are still accessible by e.g. SPARQL queries and other means. We then describe an example that illustrates the methodology and indicates the benefits of our approach.(1 2)