ON THE BEST WAY OF DESIGNATING OBJECTS IN SENTENCE GENERATION
Authors:
- 1 January 1984
- journal article
- Published by Emerald Publishing in Kybernetes
- Vol. 13 (1) , 43-46
Abstract
This paper deals with a problem encountered in natural language generation which seems to have been largely ignored in the literature, that of generating non‐ambiguous (i.e. discriminating) designations of objects in a given context, from a knowledge basis, which associates the properties and relations, concerning the objects present in the environment, with their respective formal labels. A search algorithm of type A is proposed, which always generates a discriminating designation when such a designation exists in terms of the available knowledge; for the evaluation the algorithm uses a subjective length function which takes into account the “intelligibility” of the designation. This work takes place in the SYROCO system, a dialogue interface for limited domains of discourse; the sentence interpretation as well as the sentence generation in SYROCO are briefly presented in the first part of this paper.Keywords
This publication has 3 references indexed in Scilit:
- FUZZY PATTERN MATCHINGKybernetes, 1982
- CONCEPTUAL GENERATION1975
- Understanding natural languageCognitive Psychology, 1972