Incorporating Temporal Information in Entailment Graph Mining
This addresses a specific issue in natural language processing for the sports domain, though it appears incremental as it builds on existing entailment graph methods by adding temporal constraints.
The paper tackles the problem of spurious entailments in entailment graphs by incorporating temporal information, focusing on sports events where teams play multiple times with different outcomes. The unsupervised model, evaluated on a manually constructed dataset, shows that using time intervals and temporal windows effectively learns correct entailments like win/lose → play while avoiding incorrect ones like win → lose.
We present a novel method for injecting temporality into entailment graphs to address the problem of spurious entailments, which may arise from similar but temporally distinct events involving the same pair of entities. We focus on the sports domain in which the same pairs of teams play on different occasions, with different outcomes. We present an unsupervised model that aims to learn entailments such as win/lose $\rightarrow$ play, while avoiding the pitfall of learning non-entailments such as win $\not\rightarrow$ lose. We evaluate our model on a manually constructed dataset, showing that incorporating time intervals and applying a temporal window around them, are effective strategies.