Position Statements

Kevin Ashley

Project Number: NSF IRI-9619713

Title: Adding Domain Knowledge to Inductive Learning Methods for Classifying Texts

  1. What types of IDM-related research activities that NSF could/should support (e.g., industrial collaboration, summer courses for inter-disciplinary training, industrial research, infrastructure research, etc)? (Your answers here are needed for the breakout groups #2, #3, and #4. Please answer briefly.)

Potentially, there is a great deal of interest in automated textual indexing and Textual CBR among legal publishers like Thompson (West) and Elsevier (Lexis). My efforts to obtain complementary funding from West Group have not met with success, however, despite the clear relevance of my NSF/IDM-funded work to their efforts. Perhaps NSF/IDM could facilitate obtaining some type of matching funds contributions from the legal database/publishing industry. For instance, it could provide some significant recognition for corporate participants who provide matching funds.

(5) What are the types of activities that the IDM research community can do (e.g., in conferences, in professional societies)? Please answer briefly.

At AAAI-98, I chaired a workshop on Textual Case-Based Reasoning. The workshop focused on new techniques for supporting reasoning with textual cases. This is important in developing Help-Desks and in dealing with domains like law, ethics, policy-making, and others where cases are naturally expressed in textual form. Although the workshop organizers attempted to foster participation by members of the Information Retrieval communities, for a variety of reasons the attempt failed. There was a great deal of interest in the IR community however. Perhaps the IDM research community could assist in bringing these to communities of researchers together.