Unsupervised Opinion Summarization Using Approximate Geodesics

TL;DR

This paper introduces GeoSumm, an unsupervised extractive opinion summarization system that uses approximate geodesic distances and representation learning to identify and summarize popular opinions from user reviews.

Contribution

The paper presents GeoSumm, a novel unsupervised model combining representation learning and geodesic distance scoring for opinion summarization, achieving state-of-the-art results.

Findings

GeoSumm outperforms existing methods on three datasets.

The model generalizes well across different domains.

Representation learning enhances relevance scoring.

Abstract

Opinion summarization is the task of creating summaries capturing popular opinions from user reviews. In this paper, we introduce Geodesic Summarizer (GeoSumm), a novel system to perform unsupervised extractive opinion summarization. GeoSumm involves an encoder-decoder based representation learning model, that generates representations of text as a distribution over latent semantic units. GeoSumm generates these representations by performing dictionary learning over pre-trained text representations at multiple decoder layers. We then use these representations to quantify the relevance of review sentences using a novel approximate geodesic distance based scoring mechanism. We use the relevance scores to identify popular opinions in order to compose general and aspect-specific summaries. Our proposed model, GeoSumm, achieves state-of-the-art performance on three opinion summarization datasets. We perform additional experiments to analyze the functioning of our model and showcase the generalization ability of {\X} across different domains.