Cost-sensitive regression-based recommender system

Collaborative filtering aims to predict the preferences of an active user from a database of available user preferences. These preferences are typically expressed as numerical ratings. However, existing recommender systems seldom suggest the appropriate recommendation with the predicted numerical ratings. In this paper, we propose a framework integrating the regression-based approach and the cost-sensitive learning to address this issue. Firstly, we employ the memory-based regression approach for binary recommendations. Secondly, we consider misclassification cost for determining the recommender behavior. Experimental results obtained on the well-known MovieLens data set show that the regression-based approach and the cost-sensitive learning are valid in computing the optimal recommender threshold.