Recommendation using neighborhood methods with preference-relation-based similarity

Selecting appropriate items from a list consisting of a large number of items provided by a product website can be difficult and time-consuming for the potential customers. The development of recommender systems should be an important solution that will help users to select items easily according to their preferences. For recommender systems, the main aim of the popular collaborative filtering approaches is to recommend items that users with similar preferences have liked in the past. Because there is a certain degree to which one alternative is not worse than another in decision making, it would be interesting to make further use of the preference relation to design a similarity measure by measuring the overall strength of one user’s preference over that of another. The proposed similarity of one user to another user is therefore dependent on the strength of the preference of the former over the latter. In contrast to traditional similarity measures for neighborhood methods in collaborative filtering, the proposed preference-relation-based similarity is not symmetric for any two users. Experimental results have demonstrated that the generalization ability of the proposed multi-criteria neighborhood method performs well in comparison to other single-criterion and multi-criteria neighborhood methods.