Combining User Contexts and User Opinions for Restaurant Recommendation in Mobile Environment

Abstract

In a mobile setting, user preferences vary in different contexts. Advances in mobile technologies have made the collection of user context information feasible, and as a result, the context-aware mobile recommender system field has been formed. Although there exist several different approaches to incorporating context into the recommendation process, context-aware recommendations are still difficult to compute. It is unclear which contextual factors are important and to which degree they influence user-item selection decisions. In this paper, the authors design a novel mobile restaurant recommender system combining user contexts and user opinions to provide restaurants to mobile users. This system extracts restaurant features from online user reviews and calculates the polarity of them based on sentiment analysis. It takes a new approach for assessing and modeling the relationship between contextual factors and restaurant features. On this basis, a hybrid recommender method is constructed, which integrates the contextual matching algorithm based on analytic hierarchy process and the collaborative filtering algorithm based on context similarity. From the user study, this system suggests restaurants that make the user more satisfied than another comparative system.