One for All: Towards Language Independent Named Entity Linking

TL;DR

This paper presents LIEL, a language-independent entity linking system that, once trained on English, effectively links entities across multiple languages without modification, outperforming existing systems.

Contribution

Introduces LIEL, a novel language-independent entity linking framework that generalizes across languages using global prediction and domain-independent features.

Findings

LIEL outperforms state-of-the-art English systems by 4 points.

LIEL achieves 14 points better results on Spanish datasets.

The approach demonstrates strong cross-language generalization.

Abstract

Entity linking (EL) is the task of disambiguating mentions in text by associating them with entries in a predefined database of mentions (persons, organizations, etc). Most previous EL research has focused mainly on one language, English, with less attention being paid to other languages, such as Spanish or Chinese. In this paper, we introduce LIEL, a Language Independent Entity Linking system, which provides an EL framework which, once trained on one language, works remarkably well on a number of different languages without change. LIEL makes a joint global prediction over the entire document, employing a discriminative reranking framework with many domain and language-independent feature functions. Experiments on numerous benchmark datasets, show that the proposed system, once trained on one language, English, outperforms several state-of-the-art systems in English (by 4 points) and the trained model also works very well on Spanish (14 points better than a competitor system), demonstrating the viability of the approach.