The Application of Theorem Proving to Information Retrieval
by Coles, L. Stephen
Technical Note 51
Institution: AI Center, SRI International
Address: 333 Ravenswood Ave, Menlo Park, CA 94025
Note: SRI Projects 7963 and 8696. The research reported herein was sponsored in part by the Lister Hill National Center for Biomedical Communications of the National Library of Medicine under Contract NLM-69-13, and in part by the National Science Foundation under Grant GN-895. Presented at the 5th Hawaii International Conference on Systems Sciences, Honolulu, Hawaii, 1972.
Most conventional computer information-retrieval systems are limited by rigid data structures and inflexible query languages. Computer question-answering systems designed to overcome either or both of these limitations have been built, but for the most part they have been restricted to small data bases. In this paper we will describe an approach to combining and extending recently developed question-answering techniques to reasonably large data files. A compilation of widely used physical laws and effects of interest to both engineers and scientists consisting of 105 basic data items will be used as a basis for demonstrating theorem-proving techniques on a large file. A restricted natural-language input for querying the file is also described.
|Coles, L. Stephen||Alumnus|