Recommendation algorithm based on user attributes and tag preferences

Abstract Aiming at the problem of sparse data and low recommendation accuracy of recommendation systems in a big data environment, a UT-CF algorithm that combines user attributes and tag preferences is proposed. The algorithm extracts and selects user attribute information, calculates the user attribute similarity matrix according to the user attribute characteristics; obtains the user’s tag score according to each score of the user item score matrix, obtains the user tag preference matrix, and calculates the difference between user tags Similarity, to get the user’s similar neighbors; In order to reduce the search space of similar users, the K-means clustering algorithm based on the principle of maximum distance is introduced. Make the top k recommendations based on the recommendations of the target user’s nearest neighbors. Through the experiment of MovieLens 100K data set, the algorithm improves the accuracy of recommendation.