Integrating case-based and rule-based reasoning in knowledge-based systems development

Case-based reasoning (CBR) and rule-based reasoning (RBR) are two paradigms for building knowledge-based systems. They represent distinct approaches to knowledge-based systems development and distinct cognitive models of human problem solving. To date, they have been viewed as competing, rather than complementary, paradigms. This research shows that combining CBR with RBR leads to a stronger approach to knowledge-based systems development.; The research approach is to construct, compare and contrast two expert systems, one case-based and one rule-based, to perform the same task. While claims have been made as to the relative advantages of each approach, this is the first systematic comparison of independent CBR and RBR systems built in the same domain. Strengths and weaknesses of each system are identified, and the best of both systems are combined in a hybrid system.; The domain of study is nutritional menu planning. This domain: (1) presents an AI challenge. Human experts consistently outperform computer systems in planning nutritious and appetizing menus. (2) poses a difficult problem. Unsuccessful attempts to build computer-assisted menu planners date back thirty years. (3) has supportive experts available to assist with system construction and evaluation.; Contributions of the dissertation are: (1) a new approach to CBR/RBR hybridization, in which cases contribute toward constraint satisfaction and rules contribute toward achievement of personal preference goals; (2) a new CBR metric for identifying and retrieving reusable cases, based on ease of adaptation; (3) a public domain nutritional menu planning system, accessible via the World Wide Web; (4) a framework for building special purpose therapeutic menu planning systems for use in the prevention and treatment of disease; (5) a case base of menus meeting guidelines for sound nutrition and aesthetic standards for color, texture, temperature and taste; (6) a new adaptation strategy for adjusting serving sizes in menus; (7) a deeper understanding of the distinctions between CBR and RBR and the relative strengths and weaknesses of the two paradigms.