Etimating Movie Rating From User Rating with Demographic Information

In an expansive multi-billion international film industry, online reviews inform audiences which films may be the most appealing, determining in part the success a film or franchise. Online reviewers and even review aggregations are biased by social or financial factors; this bias clouds the value of movie reviews by prioritizing and propagating highly visible user’s ratings of films, as compared to an ideal movie review independent of the individual’s social connections. This paper seeks to create a movie rating prediction model that minimizing the social bias of online movie ratings, while providing metrics that can adjust movie satisfaction ratings according to a user’s demographic and preference; these two approaches capture an effort to comprehensively capture the social impact on movie ratings. To compensate for the influence of hyper-visible user’s views on a film, we leverage user-movie rating matrix completion to penalize the weight of similar and related users. Then, to capture information on the social networks of online movie critics, to predict a demographic’s future reviews of a film, we use social network features such as influential factors, centrality, and more. Finally, we establish a statistical inference model that builds a rating for a film theoretically isolated from social bias.