English-language processing system that learns about new domains
This English-language processing system called LDC emphasises (a) small- or medium-sized office domains, as opposed to large relational-style databases; (b) mechanisms to learn about new domains and the English to be used in discussing them; and (c) capabilities for deep semantic processing, for example where English inputs can be phrased naturally, not merely as a notational variant for complete, formal queries. LDC consists of two major components and an external retrieval module. The first component, prep, obtains information about a new domain and the language to be used in discussing it. The second, user-phase, component of LDC resembles an ordinary NL processor, but (a) most decisions are determined from the preprocessed information appearing in the data files produced by prep, and (b) the emphasis is upon the semantics of layered domains, described herein. This paper presents the motivation behind LDC, summarizes and gives examples of the behavior of prep, provides an overview of the user-phase component, and gives examples of current and projected capabilities of the system. 21 references.
- OSTI ID:
- 5179922
- Country of Publication:
- United States
- Language:
- English
Similar Records
Communicating with computers in English: the emergence of natural-language processing
Interpretation of natural-language data base queries using optimization methods