Case-Based Reasoning

David Leake and Ralph Barletta

Course Description

Case-based reasoning (CBR) has become an established part of artificial intelligence, both as a means for addressing fundamental AI problems and as a basis for fielded AI technology. This tutorial defines the fundamental principles of CBR and shows how to apply them to real-world problems.

We start by presenting an overview of the case-based reasoning process, comparing and contrasting CBR to other AI methods to clarify motivations for the case-based approach. We then highlight key issues for CBR systems, principles for addressing them, and their role in high-impact applications and research systems. This discussion will be based on case studies of important CBR systems that reason and learn in multiple task domains. For each case study we talk about fundamental CBR issues the system addresses, consider how effective the system is in accomplishing its stated goal, and finally assess the contributions of the system to the state of the art in CBR.

Following this overview, we look at the major issues in designing and maintaining real-world CBR applications. We will define a methodology for building successful CBR systems in the real world, addressing tasks ranging from data collection and knowledge engineering to project scoping and user interface design. The tutorial closes by highlighting lessons learned from CBR research and applications, current challenges for case- based reasoning methods, how the challenges are being addressed, and opportunities for the next generation of CBR systems and their future impact.

The tutorial will clarify key principles of CBR, will supply applications developers with the information they need to begin selecting and applying CBR tools, and will provide experienced professionals with an in-depth view of key issues, methods, and emerging opportunities for applying case-based reasoning technology.

Prerequisite Knowledge

The tutorial is aimed at both researchers and application builders who are interested in how to integrate CBR into the systems they are building. Prior experience with CBR will certainly enhance the value of the tutorial but is definitely not required.

About the Lecturers

David Leake is an Assistant Professor of Computer Science at Indiana University. His primary research area is case-based reasoning, in which he has over 50 publications including the book "Case-Based Reasoning: Experiences, Lessons, and Future Directions" (AAAI Press, 1996). He chaired the AAAI-93 Workshop on Case-Based Reasoning, presented the opening tutorial at the First International Conference on Case-Based Reasoning in 1995, and is co-chair of the 1997 Second International Conference on Case-Based Reasoning. He received his Ph.D. in Computer Science from Yale University.

Ralph Barletta is Chief Scientist at Inference Corporation (USA), a leading vendor of software tools for problem resolution using CBR technology, where he defines and oversees research and development for CBR products. He was President of Case Data Solutions Inc., a consulting firm specializing in AI and CBR, and oversaw development of ReMind, one of the first commercial CBR tools. His numerous publications include the book "A Review Of Industrial CBR Tools," which he co-authored in 1995. He presented the opening tutorial at the Second European Workshop on Case-Based Reasoning in 1994. He received his B.A. in Business Administration and M.S. in Computer Science from Rutgers University.

Last modified: Thu Feb 20 13:39:07 JST 1997