Coreferential Reasoning Learning for Language Representation

TL;DR

CorefBERT is a new language representation model designed to explicitly capture coreferential relations, leading to improved performance on NLP tasks requiring coreferential reasoning without sacrificing overall effectiveness.

Contribution

The paper introduces CorefBERT, a novel model that explicitly models coreference, enhancing discourse understanding in language representations.

Findings

CorefBERT outperforms baselines on coreferential reasoning tasks.

Maintains comparable performance on general NLP tasks.

Source code and experiments are publicly available.

Abstract

Language representation models such as BERT could effectively capture contextual semantic information from plain text, and have been proved to achieve promising results in lots of downstream NLP tasks with appropriate fine-tuning. However, most existing language representation models cannot explicitly handle coreference, which is essential to the coherent understanding of the whole discourse. To address this issue, we present CorefBERT, a novel language representation model that can capture the coreferential relations in context. The experimental results show that, compared with existing baseline models, CorefBERT can achieve significant improvements consistently on various downstream NLP tasks that require coreferential reasoning, while maintaining comparable performance to previous models on other common NLP tasks. The source code and experiment details of this paper can be obtained from https://github.com/thunlp/CorefBERT.