Sentence Representations via Gaussian Embedding

TL;DR

GaussCSE introduces a Gaussian distribution-based contrastive learning framework for sentence embedding, capturing asymmetric relationships and entailment directions, thus enhancing the expressiveness beyond traditional point embeddings.

Contribution

This paper presents GaussCSE, a novel method that models sentence representations as Gaussian distributions to handle asymmetric relations and entailment directions.

Findings

Achieves comparable performance to existing methods on natural language inference tasks.

Effectively estimates the direction of entailment relations.

Models asymmetric relationships between sentences.

Abstract

Recent progress in sentence embedding, which represents the meaning of a sentence as a point in a vector space, has achieved high performance on tasks such as a semantic textual similarity (STS) task. However, sentence representations as a point in a vector space can express only a part of the diverse information that sentences have, such as asymmetrical relationships between sentences. This paper proposes GaussCSE, a Gaussian distribution-based contrastive learning framework for sentence embedding that can handle asymmetric relationships between sentences, along with a similarity measure for identifying inclusion relations. Our experiments show that GaussCSE achieves the same performance as previous methods in natural language inference tasks, and is able to estimate the direction of entailment relations, which is difficult with point representations.