• •    

一种基于标签权重的协同过滤推荐算法

雷曼,龚琴,王纪超,王保群   

  1. 重庆邮电大学
  • 收稿日期:2018-07-23 修回日期:2018-09-17 发布日期:2018-09-17
  • 通讯作者: 雷曼

A Collaborative Filtering Recommendation Algorithm Based on Label Weight

  • Received:2018-07-23 Revised:2018-09-17 Online:2018-09-17

摘要: 摘 要: 针对传统协同过滤推荐算法中由于相似度计算导致推荐精度不足的问题,提出一种基于标签权重相似度量方法的协同过滤推荐算法。该方法首先通过改进当前算法中标签权重的计算,并构成用户-标签权重矩阵和物品-标签权重矩阵。其次考虑到推荐系统是以用户为中心进行推荐,继而通过构建用户-物品关联矩阵来获取用户对物品最准确的评价和需求。最后根据用户-物品的二部图,利用物质扩散算法计算基于标签权重的用户间相似度,并为目标用户生成推荐列表。实验结果表明,与UITGCF的方法相比,在稀疏度环境为0.1时该算法的召回率、准确率和F1值分别提高了14.69%、9.44%、17.23%。当推荐项目数量N=10时,三个指标分别提高了17.99%、8.98%、16.27%。结果表明基于标签权重的协同过滤推荐算法有效提高推荐结果。

关键词: 用户-标签权重, 物品-标签权重, 推荐系统, 协同过滤, 物质扩散

Abstract: Abstract: Aiming at the problem that recommendation accuracy is insufficient due to the similarity calculation in the traditional collaborative filtering recommendation algorithm, a collaborative filtering recommendation algorithm based on the similarity measurement method of label weight was proposed.Firstly, the calculation of the label weights in the current algorithm was improved by the method, and a user-tag weight matrix and an item-tag weight matrix were constituted.Secondly, it is considered that the recommendation system is based on the user-centered recommendation, and then the most accurate evaluation and demand of the user was obtained through the user-item association matrix.Finally, according to the user-item bipartite graph, the similarity between users based on the label weight was calculated by the material diffusion algorithm.and a recommendation list is generated for the target user. The experimental results show that compared with the UITGCF method, when the sparsity environment is 0.1. the recall, precision, F1 of the algorithm are respectively increased by 14.69%, 9.44% and 17.23%. When the recommendation item number was N=10, the three indicators respectively increased by 17.99%, 8.98%, and 16.27%. The results show that the collaborative filtering recommendation algorithm based on tag weight effectively improves the recommendation result.

Key words: user-tag weight, item-tag weight, recommendation system, collaborative filtering, material diffusion