Cross-Lingual Fine-Grained Entity Typing

TL;DR

This paper introduces a cross-lingual fine-grained entity typing model that generalizes to over 100 languages, including unseen ones, and demonstrates superior performance over baselines in predicting entity types across languages and entities.

Contribution

The paper presents a unified model capable of handling over 100 languages for entity typing, with a focus on generalization to unseen languages and entities, outperforming simple baselines.

Findings

Outperforms baselines on unseen languages like Japanese, Tamil, Arabic, Serbian, and Persian.

Significantly improves entity type prediction for unseen entities, even in unseen languages.

Human evaluation confirms strong relevance of predicted types in challenging settings.

Abstract

The growth of cross-lingual pre-trained models has enabled NLP tools to rapidly generalize to new languages. While these models have been applied to tasks involving entities, their ability to explicitly predict typological features of these entities across languages has not been established. In this paper, we present a unified cross-lingual fine-grained entity typing model capable of handling over 100 languages and analyze this model's ability to generalize to languages and entities unseen during training. We train this model on cross-lingual training data collected from Wikipedia hyperlinks in multiple languages (training languages). During inference, our model takes an entity mention and context in a particular language (test language, possibly not in the training languages) and predicts fine-grained types for that entity. Generalizing to new languages and unseen entities are the fundamental challenges of this entity typing setup, so we focus our evaluation on these settings and compare against simple yet powerful string match baselines. Experimental results show that our approach outperforms the baselines on unseen languages such as Japanese, Tamil, Arabic, Serbian, and Persian. In addition, our approach substantially improves performance on unseen entities (even in unseen languages) over the baselines, and human evaluation shows a strong ability to predict relevant types in these settings.