Estimating probabilities in recommendation systems

摘要

Recommendation systems are emerging as an important business application with significant economic impact. Currently popular systems include Amazon's book recommenda- tions, Netflix's movie recommendations and Pandora's music recommendations. We address the problem of estimating probabilities associated with recommendation system data by using non-parametric kernel smoothing. In our estimation we interpret missing items as randomly cen-sored observations of preference relations and obtain efficient computation schemes by using combinatorial properties of generating functions. We demonstrate our approach with several case-studies involving real world movie recommendation data. The results are comparable with state of the art techniques while also providing probabilistic preference estimates outside the scope of traditional recommender systems.