Aspect and sentiment unification model for online review analysis

TLDR

User-generated reviews on the Web contain sentiments about detailed aspects of products and services, yet most are plain text, making it laborious to extract relevant details. The study aims to automatically discover which aspects are evaluated in reviews and how sentiments for different aspects are expressed. We propose Sentence‑LDA, a generative model that assumes all words in a sentence come from one aspect, and extend it to the Aspect and Sentiment Unification Model (ASUM) that jointly models aspects and sentiments to discover senti‑aspects, applying both to reviews of electronic devices and restaurants. ASUM’s discovered senti‑aspects capture key aspect‑sentiment couplings, its sentiment classification outperforms other generative models and rivals supervised methods, and it achieves these results without requiring sentiment labels.

Abstract

User-generated reviews on the Web contain sentiments about detailed aspects of products and services. However, most of the reviews are plain text and thus require much effort to obtain information about relevant details. In this paper, we tackle the problem of automatically discovering what aspects are evaluated in reviews and how sentiments for different aspects are expressed. We first propose Sentence-LDA (SLDA), a probabilistic generative model that assumes all words in a single sentence are generated from one aspect. We then extend SLDA to Aspect and Sentiment Unification Model (ASUM), which incorporates aspect and sentiment together to model sentiments toward different aspects. ASUM discovers pairs of {aspect, sentiment} which we call senti-aspects. We applied SLDA and ASUM to reviews of electronic devices and restaurants. The results show that the aspects discovered by SLDA match evaluative details of the reviews, and the senti-aspects found by ASUM capture important aspects that are closely coupled with a sentiment. The results of sentiment classification show that ASUM outperforms other generative models and comes close to supervised classification methods. One important advantage of ASUM is that it does not require any sentiment labels of the reviews, which are often expensive to obtain.