Question Answering with Imperfect Temporal Information

A temporal question answering system must be able to deduce which qualitative temporal relation holds between two events, a reasoning task that is complicated by the fact that historical events tend to have a gradual beginning and ending. In this paper, we introduce an algebra of temporal relations that is well–suited to represent the qualitative temporal information we have at our disposal. We provide a practical algorithm for deducing new temporal knowledge, and show how this can be used to answer questions that require several pieces of qualitative and quantitative temporal information to be combined. Finally, we propose a heuristic technique to cope with inconsistencies that may arise when integrating qualitative and quantitative information.