Accession Number:

ADA196719

Title:

Automatic Concept Formation in a Rich Input Domain

Descriptive Note:

Final rept. Mar 1985-Mar 1987

Corporate Author:

COLUMBIA UNIV NEW YORK DEPT OF COMPUTER SCIENCE

Personal Author(s):

Report Date:

1988-06-01

Pagination or Media Count:

49.0

Abstract:

Learning by observation involves the creation of categories summarizing experience. In this research note, we summarize our research during the contact period with UNIMEM, an Artificial Intelligence system that learns by observation. UNIMEM is a robust computer program that can be run on many domains with real-world problem characteristics, such as uncertainty, incompleteness, and large numbers of examples. We give an overview of the program that illustrates UNIMEMs key elements, including the automatic creation of non- disjoint concept hierarchies that are evaluated over time. We then describe several experiments that we have carried out with UNIMEM, testing it on different domains universities, Congressional voting records, and terrorist events, and an examination of the effect of varying UNIMEMs parameters on the resulting concept hierarchies. Finally, we discuss future directions for our work with the program. Keywords Expert systems, Reasoning, UNIMEMUniversal Memory.

Subject Categories:

  • Computer Programming and Software

Distribution Statement:

APPROVED FOR PUBLIC RELEASE