EASE: Entity-Aware Contrastive Learning of Sentence Embedding

TL;DR

EASE introduces a contrastive learning approach for sentence embeddings using entity supervision, enhancing semantic understanding and cross-lingual alignment, and demonstrates superior performance in multilingual tasks.

Contribution

The paper proposes EASE, a novel entity-aware contrastive learning method that improves sentence embeddings by leveraging entity information for better semantic and cross-lingual representations.

Findings

EASE outperforms baseline models in multilingual tasks.

EASE achieves competitive results in English semantic similarity.

Entity supervision enhances sentence embedding quality.

Abstract

We present EASE, a novel method for learning sentence embeddings via contrastive learning between sentences and their related entities. The advantage of using entity supervision is twofold: (1) entities have been shown to be a strong indicator of text semantics and thus should provide rich training signals for sentence embeddings; (2) entities are defined independently of languages and thus offer useful cross-lingual alignment supervision. We evaluate EASE against other unsupervised models both in monolingual and multilingual settings. We show that EASE exhibits competitive or better performance in English semantic textual similarity (STS) and short text clustering (STC) tasks and it significantly outperforms baseline methods in multilingual settings on a variety of tasks. Our source code, pre-trained models, and newly constructed multilingual STC dataset are available at https://github.com/studio-ousia/ease.