Abstract: In recent years, many collaborative filtering-based recommender systems have been successfully applied, but with the increasing number of system users and projects, the amount of similarity calculation increases sharply, collaborative filtering recommendation system scalability issues become increasingly prominent. This paper puts forward a LSH/MinHash algorithm based on the approximate nearest neighbor, and applies it to the clustering of library resources, for solving the problem of high dimension and a amount of data cluster in the complexity under reasonable time. It reduces the amount of similarity calculation, improves the scalability of the algorithm. Experiments show that this algorithm is of higher efficiency and accuracy.


Key words: library, personalized recommendation, collaborative filtering, LSH