A course on the knowledge representation problem in AI, with an emphasis on the use of logical techniques. One foundation of the course is ``computational logic,'' as represented, for example, by the current work on logic programming and its extensions. The course considers various modal logics that have been proposed as representations of our ``common sense categories'': e.g., time, action, knowledge, belief. Both the semantics and the proof theory of these logics will be studied, and the feasibilty of selectively incorporating modal proof procedures into an automated reasoning system will be investigated. Finally, the course will consider several forms of nondeductive inference that have been analyzed with the use of logical techniques, such as nonmonotonic reasoning and abductive reasoning. The extent to which these ``common sense inferences'' can be accommodated within a logical framework will be a major point of discussion.
Additional topics, possibly dealing with the complexity of reasoning, may be added at the discretion of the instructor.
Weekly reading assignments, several problem sets, and a final examination.