Recommendation systems have wide-spread applications in both academia andindustry. Traditionally, performance of recommendation systems has beenmeasured by their precision. By introducing novelty and diversity as keyqualities in recommender systems, recently increasing attention has beenfocused on this topic. Precision and novelty of recommendation are not in thesame direction, and practical systems should make a trade-off between these twoquantities. Thus, it is an important feature of a recommender system to make itpossible to adjust diversity and accuracy of the recommendations by tuning themodel. In this paper, we introduce a probabilistic structure to resolve thediversity-accuracy dilemma in recommender systems. We propose a hybrid modelwith adjustable level of diversity and precision such that one can perform thisby tuning a single parameter. The proposed recommendation model consists of twomodels: one for maximization of the accuracy and the other one forspecification of the recommendation list to tastes of users. Our experiments ontwo real datasets show the functionality of the model in resolvingaccuracy-diversity dilemma and outperformance of the model over other classicmodels. The proposed method could be extensively applied to real commercialsystems due to its low computational complexity and significant performance.
展开▼
