Abstract: | In order for human-computer interaction to be effective, not only must a computer system be able to understand a large subset of natural language, but it must also be able to communicate with the user in a natural way. A response by a system to a query must be cooperative, informative and non-misleading, in order to imitate human question-answering behaviour.The aim of this project was to implement a response module in an existing NLIDB, HPSG-QAS I, that generates HPSG feature structures for a range of English questions, transforms them into appropriate logical queries and generates answers by evaluating the queries against a simple Prolog database.
|