Blindness to Modality Helps Entailment Graph Mining

TL;DR

This paper demonstrates that removing modal modifiers from predicates in entailment graph learning can improve performance, highlighting the complex role of modality in linguistic inference tasks.

Contribution

It introduces a novel approach of filtering modal modifiers in entailment graph construction, revealing that ignoring modality can enhance entailment detection.

Findings

Stripping modal modifiers increases entailment graph performance.

Modal pragmatics can sometimes hinder rather than help inference.

Filtering modality is a useful preprocessing step for entailment tasks.

Abstract

Understanding linguistic modality is widely seen as important for downstream tasks such as Question Answering and Knowledge Graph Population. Entailment Graph learning might also be expected to benefit from attention to modality. We build Entailment Graphs using a news corpus filtered with a modality parser, and show that stripping modal modifiers from predicates in fact increases performance. This suggests that for some tasks, the pragmatics of modal modification of predicates allows them to contribute as evidence of entailment.