Temporal Relation Classification Based on Temporal Reasoning
Francisco Costa, António Branco · 2013
The area of temporal information extraction has recently focused on temporal relation classification. This task is about classifying the temporal relation (precedence, overlap, etc.) holding between two given entities (events, dates or times) mentioned in a text. This interest has largely been driven by the two recent TempEval competitions. Even though logical constraints on the structure of possible sets of temporal relations are obvious, this sort of information deserves more exploration in the context of temporal relation classification. In this paper, we show that logical inference can be used to improve—sometimes dramatically— existing machine learned classifiers for the problem of temporal relation classification. 1