An English language question answering system for a large relational database
Communications of the ACM
Hi-index | 0.02 |
Compared to humans, current natural language dialogue systems often behave in a rigid and fragile manner when their conversations deviate from a narrowly conceived mainstream, e.g. when faced with ungrammatical, unclear, or unrecognizable input, ambiguous descriptions, or requests for clarification of their own output. We believe that the time is now ripe to construct systems which can interact gracefully with their users when such contingencies arise. Graceful interaction is not a single skill, but a combination of several diverse abilities. We list these components, and describe one of them - the ability to communicate robustly. Detailed descriptions of all the components appear in [4], along with details of a system architecture for their integrated Implementation.