Today, a large volume of hotel reviews is available on many websites, such as TripAdvisor (http://www.tripadvisor.com) and Orbitz (http://www.orbitz.com). A typical review contains an overall rating and several aspect ratings along with text. The rating is perceived as an abstraction of reviewers’ satisfaction in terms of points. Although the amount of reviews having aspect ratings is growing, there are plenty of reviews including only an overall rating. Extracting aspect-specific opinions hidden in these reviews can help users quickly digest them without actually reading through them. The task mainly consists of two parts: aspect identification and rating inference. Most existing studies cannot utilize aspect ratings which are becoming abundant in the last few years. In this paper, we propose two topic models which explicitly model aspect ratings as observed variables to improve the performance of aspect rating inference over unrated reviews. Specifically, we consider sentiment distributions in the aspect level, which generate sentiment words and aspect ratings. The experiment results show our approaches outperform other existing methods on the data set crawled from Trip Advisor.
