Sentiment-based Topic Model for Mining Usability Issues and Failures with User Products

Abstract

This paper describes an approach to problem phrase extraction from texts that contain user experience with products. User reviews from online resources, that describe actual difficulties in use of products in addition to sentiment-oriented phrases, affect on other people's purpose decisions. In this paper we present two probabilistic graphical models which aims to extract problems with products from online reviews. We incorporate information about problem phrases with words’ sentiment polarities (negative, neutral or positive). The proposed models learn a distribution over words, associated with topics, both sentiment and problem labels. The algorithms achieve a better performance in comparison to several state-of-the-art models in terms of the likelihood of a held-out test and in terms of an accuracy of classification results, evaluated on reviews from several different domains in English and Russian. Our contribution is that summarizing sentiment and problem information about words with reviews’ topics by the model's asymmetric priors gives an improvement for problem phrase extraction from user reviews.

Keywords

• отзывы пользователей, латентное размещение дирихле, latent dirichlet allocation, совместная вероятностная модель, извлечение проблемных высказываний