Education
Chair Positions
Area
Artificial Intelligence
Interests
Development and application of representation and reasoning techniques.
Summary
My current research focuses on the development and application of non-deductive reasoning techniques (non-monotonic reasoning and belief revision) and their application to automated diagnosis, scheduling, database mining, and related areas. The challenge is to retain the clarity and robustness of good theory, while making progress in their practical deployment in real applications.
The current most active application areas are intelligent scheduling, automated layout, and data base mining. The methods for scheduling and layout are based on extensions of the theory and practice of constraint solving and constraint logic programming. In both cases, there exist important and difficult challenges with respect to the incremental specification of complex scheduling and layout constraints, and the incremental dynamic specification of optimization criteria.
Related work includes the development of constraint languages and algorithms for restricted classes of application domains. These include personnel scheduling, high school timetabling, and the parallel development of visualization and user interaction tools to help develop specifications for these problems. With respect to data mining, the underlying methods again use constraint logic programming tools, augmented with more powerful inference techniques based on various forms of induction. Application-specific induction projects include the creation of hypotheses on protein structure, and the automatic formation of semantic indexing structures used to support content-based retrieval of text and images.