Retrieval of collaborative filtering nearest neighbors in a content-addressable space

Collaborative Filtering (CF) is considered one of the popular and most widely used recommendation techniques. It is aimed at generating personalized item recommendations for the users based on the assumption that similar users have similar preferences and like similar items. One of the major drawbacks of the CF is its limited scalability, as the CF computational effort increases linearly with the number of users and items. This work presents a novel variant of the CF, employed over a content-addressable space. This heuristically decreases the computational effort required by the CF by restricting the nearest neighbors search applied by the CF to a set potentially highly similar users. Experimental evaluation demonstrates that the proposed approach is capable of generating accurate recommendations, while significantly improving the performance in comparison with the traditional implementation of the CF.