Abstract: A user gradebased collaborative filtering recommendation algorithm is presented in this paper. It solves the scalability problem of traditional collaborative filtering algorithm. It defines a user grade function firstly, which determines users’ grade according to the number of their rating items, and by means of finding neighbors for a target user, it only considers the candidates in his near grade, so the efficiency increases. Both theory and experimental results show that this algorithm improves recommendation efficiency greatly compared with traditional collaborative filtering, and at the same time the recommendation quality does not decrease.

Key words: collaborative filtering, user grade, recommendation algorithm