Neural Coreference Resolution based on Reinforcement Learning

TL;DR

This paper introduces a reinforcement learning-based neural coreference resolution system that jointly performs mention detection and clustering, achieving state-of-the-art results on the CoNLL-2012 dataset using BERT.

Contribution

It presents a novel actor-critic reinforcement learning approach for coreference resolution that integrates mention detection and clustering in a unified framework.

Findings

Achieves state-of-the-art performance on CoNLL-2012 English Test Set.

Leverages BERT span representations for improved accuracy.

Demonstrates effectiveness of reinforcement learning in coreference tasks.

Abstract

The target of a coreference resolution system is to cluster all mentions that refer to the same entity in a given context. All coreference resolution systems need to solve two subtasks; one task is to detect all of the potential mentions, and the other is to learn the linking of an antecedent for each possible mention. In this paper, we propose a reinforcement learning actor-critic-based neural coreference resolution system, which can achieve both mention detection and mention clustering by leveraging an actor-critic deep reinforcement learning technique and a joint training algorithm. We experiment on the BERT model to generate different input span representations. Our model with the BERT span representation achieves the state-of-the-art performance among the models on the CoNLL-2012 Shared Task English Test Set.