In Proceedings of the Technology Demonstration Session of the 14th International Conference on Knowledge Engineering and Knowledge Management, EKAW 2004, 5-8th October 2004 - Whittlebury Hall, Northamptonshire, UK. A Learning Agent Shell for Building Knowledge-Based Agents Gheorghe Tecuci, Mihai Boicu, Dorin Marcu, Bogdan Stanescu, Cristina Boicu, Marcel Barbulescu Learning Agents Center, Department of Computer Science MSN 4A5, George Mason University, Fairfax, VA 22030, USA Tel: 1 703 993 {1722, 1591, 1535, 1535, 4669, 1535} {tecuci, mboicu, dmarcu, bstanesc, ccascava, mbarbule}@gmu.edu ABSTRACT This paper presents Disciple-RKF, a learning agent shell that can be used by subject matter experts, with limited assistance from knowledge engineers, to develop knowledge-based agents incorporating their expertise. Keywords Knowledge acquisition, learning agent shell, tools for subject matter experts, knowledge engineering, ontology, rules. 1. DISCIPLE-RKF LEARNING AGENT SHELL Disciple-RKF is a learning agent shell that represents the most recent implementation of Disciple, an evolving theory and methodology for the development of knowledge bases and agents, by subject matter experts, with limited assistance from knowledge engineers [1, 2, 3]. A main goal of the Disciple approach is to overcome the knowledge acquisition bottleneck in the development of knowledge-based systems. The knowledge-based agents developed with Disciple-RKF use task-reduction as the main problem solving paradigm. In this paradigm, a problem solving task is successively reduced to simpler tasks, the solutions of the simplest tasks are found, and these solutions are successively composed into the solution of the initial task. The knowledge base of an agent is structured into an object ontology that represents the objects from an application domain, and a set of task reduction and solution composition rules expressed with these objects. The Disciple-RKF shell is as a general problem solving and learning agent with no specific knowledge in its knowledge base. To develop an agent for a specific application domain, one needs to develop its knowledge base by using the various modules of Disciple-RKF. In essence, this process consists of developing the ontology for the specific application domain and of teaching the agent how to perform various tasks, in a way that resembles how one would teach a human apprentice. In the last four years, successive versions of Disciple-RKF have been used to develop intelligent agents for center of gravity analysis that are used in several courses at the US Army War College. The next section presents an overview of the Disciple-RKF agent development methodology. 2. AGENT DEVELOPMENT METHODOLOGY The Disciple approach covers all the phases of agent development and use. First, a knowledge engineer works with a subject matter expert to develop an ontology for the application domain. They use the ontology import module (to extract relevant ontology elements from existing knowledge repositories) as well as the various ontology editors and browsers of Disciple-RKF. Fig. 1 shows the interfaces of three of Disciple’s ontology browsers: at top is the association browser (which displays an object and its relationships with other objects), at bottom left is the tree browser (which displays the hierarchical relationships between the objects in a tree structure), and at bottom right is the graphical browser (which displays the hierarchical relationships between the objects in a graph structure). The result of this knowledge base development phase is an object ontology which is complete enough to be used as a generalization hierarchy for learning, allowing the expert to train the Disciple agent how to solve problems, with limited assistance from a knowledge engineer. Figure 1. Three ontology browsers of Disciple-RKF European Conference on Knowledge Engineering and Knowledge Management (EKAW’04), October 5–8, 2004, Northamptonshire, UK. Copyright resides with the author(s) of this manuscript.