Experimental Methods for Evaluating Planning Systems.
Abstract
The purpose of this research was to develop empirical tools that allow AI researchers to study their systems in a principled, systematic way. In the course of a three-year project the University of Massachusetts published over 30 refereed articles on empirical methods; a textbook, Empirical Methods for Artificial Intelligence; a web site devoted to research methods in AI and Computer Science; and over 40 papers at symposia, workshops, ad related forums. This final report contains summaries of the first two years of the project and the final year. The project went according to plan with one significant deviation. The project didn't envision the power of the World Wide Web. Technical results were planned to be delivered in conventional ways by publishing software and reports. By the second year of the project, it became clear that a web site devoted to empirical methods was required. MIT Press granted permission to post excerpt of the recently published book. Technical articles were posted, along with more of the tutorial and resource material, onto the web site. The Evaluation of Intelligent Systems (EIS) web site is completed, and much of the technology developed under this program - including the CLASP system - can be found there. As the web site address is changing in the near future, we refer interested readers to the Experimental Knowledge Systems Laboratory web site (http://www-eks1.cs.umass.), from which pointers to EIS may be followed.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1998
- Accession Number
- ADA348493
Entities
People
- Adele E. Howe
- Paul R. Cohen
Organizations
- University of Massachusetts Amherst