Optimization And Implementation Of Item-based Collaborative Filtering Algorithm Based on Attributes and Penalty Factors

摘要

A new item-based collaborative filtering algorithm based on attribute similarity and Penalty was proposed by analyzing the drawbacks of traditional item-based collaborative filtering algorithm according to the similarity between items to select the nearest neighbor [1]. The new item-based collaborative filtering algorithm uses the similarity of the item attributes to modify the original item similarity calculation method, and dynamically generates the punish factors according to the item's heat. It comprehensively considers the influence of the item attributes and item heat on the recommendation system, and improves the traditional item similarity measure method. The experimental results on the Movie Lens dataset show that the proposed algorithm can effectively solve the problem of sparse evaluation data and inaccurate recommendation results [2].