# Improving Sentence Similarity Estimation for Unsupervised Extractive Summarization

## TL;DR

This paper introduces two novel strategies leveraging contrastive and mutual learning to improve sentence similarity estimation, thereby enhancing unsupervised extractive summarization by better capturing document-level context and sentence salience.

## Contribution

It proposes contrastive learning and mutual learning techniques to improve sentence similarity estimation for unsupervised summarization, addressing limitations of pre-trained language models.

## Key findings

- Enhanced sentence similarity estimation improves summarization quality.
- Strategies outperform baseline methods in experimental evaluations.
- Document-level information integration leads to better sentence ranking.

## Abstract

Unsupervised extractive summarization aims to extract salient sentences from a document as the summary without labeled data. Recent literatures mostly research how to leverage sentence similarity to rank sentences in the order of salience. However, sentence similarity estimation using pre-trained language models mostly takes little account of document-level information and has a weak correlation with sentence salience ranking. In this paper, we proposed two novel strategies to improve sentence similarity estimation for unsupervised extractive summarization. We use contrastive learning to optimize a document-level objective that sentences from the same document are more similar than those from different documents. Moreover, we use mutual learning to enhance the relationship between sentence similarity estimation and sentence salience ranking, where an extra signal amplifier is used to refine the pivotal information. Experimental results demonstrate the effectiveness of our strategies.