Multi-Multi-View Learning: Multilingual and Multi-Representation Entity Typing

TL;DR

This paper introduces a multiview learning approach utilizing language and representation views to improve entity typing in knowledge bases, demonstrating effectiveness across multilingual and fine-grained classification tasks.

Contribution

It proposes a novel multiview learning framework combining language and representation views, and releases MVET, a large multilingual dataset for entity typing evaluation.

Findings

Multiview learning improves entity typing accuracy.

The approach is effective across multiple languages and representations.

MVET dataset enables comprehensive evaluation of multilingual entity typing.

Abstract

Knowledge bases (KBs) are paramount in NLP. We employ multiview learning for increasing accuracy and coverage of entity type information in KBs. We rely on two metaviews: language and representation. For language, we consider high-resource and low-resource languages from Wikipedia. For representation, we consider representations based on the context distribution of the entity (i.e., on its embedding), on the entity's name (i.e., on its surface form) and on its description in Wikipedia. The two metaviews language and representation can be freely combined: each pair of language and representation (e.g., German embedding, English description, Spanish name) is a distinct view. Our experiments on entity typing with fine-grained classes demonstrate the effectiveness of multiview learning. We release MVET, a large multiview - and, in particular, multilingual - entity typing dataset we created. Mono- and multilingual fine-grained entity typing systems can be evaluated on this dataset.