Knowledge Representation and Default Reasoning

One of the fundamental challenges in Artificial Intelligence is how to represent and reason with knowledge and belief, and default reasoning appears to be the only tractable method to deal with the problem of incomplete information, often associated with knowledge and belief.

Default reasoning has attracted considerable attention during the last decade and several different frameworks for default reasoning have been proposed, such as default logic, circumscription, and autoepistemic logic. Among others, modal logic appears to be a near perfect mathematical tool for default reasoning because it has long been realized that modal logic presents an ideal tool for reasoning with knowledge and belief, and default reasoning is basically reasoning with knowledge and belief. In fact, many attempts to develop a unified framework for default reasoning have been reported, though the results are not satisfactory, mainly because dramatically different nonmonotonic reasoning mechanisms are used in various frameworks.

We have recently observed that default reasoning can be characterized by appropriate introspection of reasoning agents, and introspective reasoning can be characterized by classical modal logic. Therefore, we believe a unified framework for default reasoning based on classical modal logic can indeed be established.

The main objective of this research is to develop a unified framework, based on classical modal logic, for default reasoning, and investigate appropriate frameworks for implementing default reasoning systems based on the proof procedures of the proposed modal logic.

Back to the home page