Speaker: Dr. Evangelos Triantaphyllou

Title: "Data Mining & Knowledge Discovery Via a Logic-Based Approach"

place: 175 Coates Hall
Date: Wednesday, November 10, 2004
Time: 3:30pm

Abstract:

We discuss the development of a data mining and knowledge discovery approach which is based on mathematical logic. Unlike many other approaches, approaches that are based on mathematical logic identify patterns in datasets in the form of Boolean functions which in turn are easier to be validated, comprehended and used by domain experts. We also present methods for guided learning, incremental learning, and decomposition of large-scale problems among other extensions of the basic approach. Furthermore, we discuss empirical results and examples from real-world datasets. The findings suggest that a logic-based approach to data mining and knowledge discovery has many advantages and looks promising for future developments in this field.