A Clustering Method of User Preference and Distance Weighted Integrating Time Factor

Abstract

A user preference and distance-weighted clustering method that integrates time factors, and alleviates the user cold start problem by introducing a user-user attribute matrix constructed from the basic objective characteristics of users, while the sparsity problem is mainly improved by introducing item features, Since the characteristics of the item can reflect the user's preference from the aspect of content, the dimension of the matrix can be reduced; the item feature is introduced into the user-item rating to obtain a small-dimensional user-item attribute total score matrix; use TF‑IDF The algorithm introduces item features when constructing the user-item attribute preference matrix, and at the same time considers the impact of user interest drifting over time on user preference; based on the above three matrices, the weighted Euclidean distance is obtained, and then the K-Means algorithm is used for clustering. This method takes movie recommendation as an example. Experimental results on the MovieLens dataset show that this method has better recommendation quality and performance than other related algorithms.

Description

technical field [0001] The invention relates to a personalized recommendation algorithm, and specifically provides a clustering method that integrates user preferences of time factors and distance weighting. Background technique [0002] In recent years, with the development of information technology and Web2.0, the information on the Internet has experienced an unprecedented surge, and problems have followed, mainly including the problem of information overload and the problem that users cannot accurately select relevant information. The recommendation system is One of the effective tools to overcome the problem of information overload. The core of the recommendation system is to design the recommendation algorithm. Therefore, various recommendation algorithms have been proposed in academia. Currently, the main recommendation algorithms used include content-based recommendation algorithms, combination recommendation algorithms, and collaborative filtering recommendation alg...

AI Technical Summary

Problems solved by technology

[0010] The purpose of the present invention is to solve the problem of user cold start and data sparsity in the original collaborative filtering recommendation algorithm, optimize and improve the existing algorithm, and design a user preference and distance weighted algorithm that integrates time factors clustering method

Images