End-to-end Neural Coreference Resolution

TL;DR

This paper presents the first end-to-end neural model for coreference resolution that outperforms previous methods without relying on syntactic parsers or hand-engineered mention detectors, achieving state-of-the-art results.

Contribution

It introduces a novel end-to-end neural approach that directly models all spans as potential mentions and learns antecedent distributions, eliminating the need for external resources.

Findings

Achieves 1.5 F1 improvement on OntoNotes benchmark.

Outperforms previous models without external resources.

Ensemble of 5 models improves F1 by 3.1 points.

Abstract

We introduce the first end-to-end coreference resolution model and show that it significantly outperforms all previous work without using a syntactic parser or hand-engineered mention detector. The key idea is to directly consider all spans in a document as potential mentions and learn distributions over possible antecedents for each. The model computes span embeddings that combine context-dependent boundary representations with a head-finding attention mechanism. It is trained to maximize the marginal likelihood of gold antecedent spans from coreference clusters and is factored to enable aggressive pruning of potential mentions. Experiments demonstrate state-of-the-art performance, with a gain of 1.5 F1 on the OntoNotes benchmark and by 3.1 F1 using a 5-model ensemble, despite the fact that this is the first approach to be successfully trained with no external resources.