Entity Disambiguation with Entity Definitions

TL;DR

This paper explores the use of more expressive textual representations, including entity definitions, to improve entity disambiguation, achieving state-of-the-art results on multiple benchmarks and enhancing generalization to unseen patterns.

Contribution

It introduces the use of richer textual representations beyond Wikipedia titles for entity disambiguation, demonstrating improved performance and generalization.

Findings

Achieved new state-of-the-art on 2 out of 6 benchmarks.

Extractive models excel with expressive representations.

Significant improvement in handling unseen entity patterns.

Abstract

Local models have recently attained astounding performances in Entity Disambiguation (ED), with generative and extractive formulations being the most promising research directions. However, previous works limited their studies to using, as the textual representation of each candidate, only its Wikipedia title. Although certainly effective, this strategy presents a few critical issues, especially when titles are not sufficiently informative or distinguishable from one another. In this paper, we address this limitation and investigate to what extent more expressive textual representations can mitigate it. We thoroughly evaluate our approach against standard benchmarks in ED and find extractive formulations to be particularly well-suited to these representations: we report a new state of the art on 2 out of 6 benchmarks we consider and strongly improve the generalization capability over unseen patterns. We release our code, data and model checkpoints at https://github.com/SapienzaNLP/extend.