Evaluating the Impact of a Hierarchical Discourse Representation on Entity Coreference Resolution Performance

TL;DR

This paper demonstrates that incorporating hierarchical discourse parse trees into neural models significantly improves entity coreference resolution performance, especially when considering different mention types.

Contribution

It introduces a novel approach that leverages automatically constructed discourse parse trees within neural models for coreference resolution, showing notable performance gains.

Findings

Significant improvement on benchmark datasets

Impact varies with mention type

Hierarchical discourse structure enhances coreference resolution

Abstract

Recent work on entity coreference resolution (CR) follows current trends in Deep Learning applied to embeddings and relatively simple task-related features. SOTA models do not make use of hierarchical representations of discourse structure. In this work, we leverage automatically constructed discourse parse trees within a neural approach and demonstrate a significant improvement on two benchmark entity coreference-resolution datasets. We explore how the impact varies depending upon the type of mention.