Clustering Aspect-related Phrases by Leveraging Sentiment Distribution Consistency

Clustering aspect-related phrases in terms of product’s property is a precursor process to aspect-level sentiment analysis which is a central task in sentiment analysis. Most of existing methods for addressing this problem are context-based models which assume that domain synonymous phrases share similar co-occurrence contexts.
In this paper, we explore a novel idea, sentiment distribution consistency, which states that different phrases (e.g. “price”, “money”, “worth”, and “cost”) of the same aspect tend to have consistent sentiment distribution. Through formalizing sentiment distribution consistency as soft constraint, we propose a novel unsupervised model in the framework of Posterior Regularization (PR) to cluster aspectrelated phrases. Experiments demonstrate that our approach outperforms baselines remarkably.