Zero-shot Neural Transfer for Cross-lingual Entity Linking

TL;DR

This paper introduces a zero-shot cross-lingual entity linking method that uses pivot languages and phonological representations to improve linking accuracy in low-resource languages without bilingual resources.

Contribution

It proposes a novel pivot-based neural framework and phonological representations for zero-shot entity linking across multiple low-resource languages.

Findings

Improves entity linking accuracy by 17% on average over baselines.

Using phonological representations increases accuracy by 36%.

Effective transfer across 54 language pairs.

Abstract

Cross-lingual entity linking maps an entity mention in a source language to its corresponding entry in a structured knowledge base that is in a different (target) language. While previous work relies heavily on bilingual lexical resources to bridge the gap between the source and the target languages, these resources are scarce or unavailable for many low-resource languages. To address this problem, we investigate zero-shot cross-lingual entity linking, in which we assume no bilingual lexical resources are available in the source low-resource language. Specifically, we propose pivot-based entity linking, which leverages information from a high-resource "pivot" language to train character-level neural entity linking models that are transferred to the source low-resource language in a zero-shot manner. With experiments on 9 low-resource languages and transfer through a total of 54 languages, we show that our proposed pivot-based framework improves entity linking accuracy 17% (absolute) on average over the baseline systems, for the zero-shot scenario. Further, we also investigate the use of language-universal phonological representations which improves average accuracy (absolute) by 36% when transferring between languages that use different scripts.