Making sense of corrupt input: contextual interpretation in LOQUI

摘要

LOQUI is the name of a natural language interface to databasesndeveloped during the course of the ESPRIT project LOKI. LOQUI is writtennin BIM prolog. The purpose of the paper is to illustrate LOQUI'snexploitation of context in its treatment of corrupt input. This is annimportant issue in LOQUI, since the system has been designed to have anrich dialogue capability that will allow conversations to proceed withnthe fluidity that is characteristic of natural human conversation,nthereby allowing both system and user the possibility of exploitingncontextual information in order to minimise constraints onnexpressiveness. This is used not only to overcome problems of corruptninput, of course, but also to allow the system to be more than just anquestion-and-answer system. The paper considers how LOQUI handlesnreference resolution, spelling correction, semantic deviance, andnpresupposition failure