Accession Number:

ADA235015

Title:

Retrieval Using Plausible Inference.

Descriptive Note:

Final rept. 1 Jan-31 Dec 1990

Corporate Author:

MASSACHUSETTS UNIV AMHERST DEPT OF COMPUTER AND INFORMATION SCIENCE

Personal Author(s):

• Croft, W. B.

Report Date:

1991-04-04

Pagination or Media Count:

19.0

Abstract:

Retrieval is a crucial function in information systems, but the algorithms used in many systems are not effective in locating the required information. For example, text-based systems are an important class of information system where both the query specification and the stored information will be incomplete and uncertain. In this situation, simple models of retrieval based on deductive inference are not adequate. We have proposed a new computational framework that views retrieval as a process of combining multiple sources of evidence, and have carried out a number of experiments in the domain of text retrieval. The experiments show that significant retrieval effectiveness improvements are possible. We have also made progress in the area of text representation using natural language processing, which provides additional evidence for the retrieval model.

Descriptors:

• *ALGORITHMS
• INFORMATION SYSTEMS
• INFORMATION PROCESSING
• INFORMATION RETRIEVAL
• NATURAL LANGUAGE
• SOURCES
• MODELS

Subject Categories:

• Information Science

Distribution Statement:

APPROVED FOR PUBLIC RELEASE