Knowledge-aware Pronoun Coreference Resolution

TL;DR

This paper presents a neural model for pronoun coreference resolution that leverages knowledge graphs with a knowledge attention mechanism, improving accuracy across domains and outperforming existing methods.

Contribution

It introduces a knowledge-aware neural approach using triplet-formatted knowledge and a selective attention module for better coreference resolution.

Findings

Outperforms state-of-the-art baselines significantly.

Demonstrates strong cross-domain generalization.

Effectively utilizes external knowledge for improved accuracy.

Abstract

Resolving pronoun coreference requires knowledge support, especially for particular domains (e.g., medicine). In this paper, we explore how to leverage different types of knowledge to better resolve pronoun coreference with a neural model. To ensure the generalization ability of our model, we directly incorporate knowledge in the format of triplets, which is the most common format of modern knowledge graphs, instead of encoding it with features or rules as that in conventional approaches. Moreover, since not all knowledge is helpful in certain contexts, to selectively use them, we propose a knowledge attention module, which learns to select and use informative knowledge based on contexts, to enhance our model. Experimental results on two datasets from different domains prove the validity and effectiveness of our model, where it outperforms state-of-the-art baselines by a large margin. Moreover, since our model learns to use external knowledge rather than only fitting the training data, it also demonstrates superior performance to baselines in the cross-domain setting.