The role of user mood in movie recommendations

Studies in consumer research indicate that mood states have effects on user behaviors and evaluation. In movie recommendation, a user in a bad mood might decide to rate some movies more harshly. In this paper, we examine how users’ mood can have an impact on their appraisal of movies in different genres, which in turn can help inform recommender system of picking up movies that are appropriate for users in different mood. Specifically, we carried out two studies. The first consists of a series of user studies to examine user mood and movie ratings to answer questions like: will a user in a more positive mood tend to rate a romantic comedy higher? Will a user in a more nervous mood tend to rate an action movie higher? Then, drawn upon the results from the first study, we modify the traditional collaborative-filtering based recommendation approach by injecting user mood and proposed a mood-aware collaborative-filtering approach. Empirical studies demonstrate that the mood-aware recommendation approach performs better than traditional one that does not consider mood.