A natural language querying system based on discourse representation theory and incorporating event semantics

Abstract

This thesis presents a querying system that accepts natural language inputs and is able to answer natural language queries about the inputs. The system works by translating English discourses into discourse representation structures, which are then transformed into Prolog facts and queries. The program is an extension of a previous implementation of Discourse Representation Theory. Added functionality includes: • asserting facts into and querying the Prolog database. • handling the past tense. • handling progressive forms. • distinguishing between states and events. The program was developed using GULP, an extension of Quintus Prolog that facilitates the implementation of unification–based grammars.