The Collaborative Agents Technology Systems Project: Introduction and Search Agents

This is the first of two talks discussing work taking place as part of the University of Maryland's new Collaborative Agents Technology Systems (CATS) Project. In this talk, Dr. Hendler describes some of the problems that CATS is set up to attack, and describes work in knowledge bases and web agents that form some of the prototype applications being ported to the CATS framework. In a second talk, to occur in November, Dr. Subrahmanian will describe fundamental new theoretical work and experimental implementations in support of agent-based computing. In this talk, Dr. Hendler will discuss "intelligent agents" technology. One of the most exciting changes in computing in the past decade has been the introduction of the world wide web and the internationally expanding internet, which make enormous amounts of information available to users regardless of location. This access, however, has made for information management problems beyond the range of most systems to handle. AI systems, combining intelligent agent technologies with large knowledge bases have long been proposed as a leading contender for helping to deal with searching, managing, and filtering this wealth of knowledge. Unfortunately, to date AI systems have not yet accomplished all that is expected from them in the information technology arena. One of the reasons for this is the lack of a memory infrastructure that can support such searches. Traditional databases don't provide enough semantics or expressivity for the needs of information agents, while AI KR search agents. In this talk we look at some work in High Performance AI that focuses on the development of large knowledge bases. We describe the PARKA-DB system, a knowledge representation technology designed for supporting efficient inferencing in frame- and description-logic-based systems containing millions of assertions. We describe several applications: including web agents, datamining, and on-line thesauri . --------------------------------------------------------------------------- Biography: Dr. Hendler is an associate professor in the Department of Computer Science, The Institute for Systems Research, and the UM Institute for Advanced Computer Studies at the University of Maryland, where he has worked since January, 1986. He is the author of the book "Integrating Marker-Passing and Problem Solving: An activation spreading approach to improved choice in planning" and is the editor of "Expert Systems: The User Interface," "Readings in Planning" (with J. Allen and A. Tate), and "Massively Parallel AI" (with H. Kitano). Hendler also serves as the Artificial Intelligence area editor for the international journal "Connection Science," is an associate editor of the "Journal of Experimental and Theoretical AI," and is on the editorial boards of "Autonomous Robots" and "IEEE Expert." Dr. Hendler was the recipient of a 1995 Fulbright Fellowship for his work in artificial intelligence. ------------------------------------------------------------------------