Extracting Aspects Hierarchies using Rhetorical Structure Theory

TL;DR

This paper introduces an unsupervised method leveraging Rhetorical Structure Theory and graph analysis to generate accurate aspect hierarchies from large review datasets, enhancing sentiment analysis at multiple levels.

Contribution

It presents a novel unsupervised approach for extracting aspect hierarchies that outperforms previous methods and can be extended with sentiment analysis.

Findings

Achieved 80% coverage compared to human hierarchies

Validated on 100,000 Amazon reviews

Easily extendable with sentiment analysis models

Abstract

We propose a novel approach to generate aspect hierarchies that proved to be consistently correct compared with human-generated hierarchies. We present an unsupervised technique using Rhetorical Structure Theory and graph analysis. We evaluated our approach based on 100,000 reviews from Amazon and achieved an astonishing 80% coverage compared with human-generated hierarchies coded in ConceptNet. The method could be easily extended with a sentiment analysis model and used to describe sentiment on different levels of aspect granularity. Hence, besides the flat aspect structure, we can differentiate between aspects and describe if the charging aspect is related to battery or price.