Joseph Phillips, University of Michigan
The proliferation of sensors and the ease of large dataset maintenance have given many scientists more data than can be analyzed by traditional means. Computers have long been used to help scientists calculate. Science, however, is more than calculating. Science at least also involves hypothesis generation and testing, planning, and integrating prior knowledge with new ideas. Artificial intelligence (AI) and database (DB) technologies have grown to the point where they may be able to help scientists in more meaningful ways. We investigate a principled approach to semi-automated knowledge discovery in databases (KDD) for integrated scientific discovery and the rationale for this approach.