Contextual Augmentation for Entity Linking using Large Language Models

TL;DR

This paper introduces a unified model leveraging large language models to improve entity linking by jointly recognizing and disambiguating entities, achieving state-of-the-art results especially on out-of-domain datasets.

Contribution

It presents a novel fine-tuned, unified framework that uses large language models for contextual augmentation in entity linking, enhancing disambiguation accuracy.

Findings

Achieves state-of-the-art performance on benchmark datasets.

Outperforms baseline models on out-of-domain data.

Demonstrates the effectiveness of contextual augmentation.

Abstract

Entity Linking involves detecting and linking entity mentions in natural language texts to a knowledge graph. Traditional methods use a two-step process with separate models for entity recognition and disambiguation, which can be computationally intensive and less effective. We propose a fine-tuned model that jointly integrates entity recognition and disambiguation in a unified framework. Furthermore, our approach leverages large language models to enrich the context of entity mentions, yielding better performance in entity disambiguation. We evaluated our approach on benchmark datasets and compared with several baselines. The evaluation results show that our approach achieves state-of-the-art performance on out-of-domain datasets.