融合兴趣差异和评分差异的协同过滤推荐算法  

Collaborative filtering recommendation algorithm combining interest difference and rating difference

作　　者：姬璐 于万钧 陈颖 JI Lu;YU Wan-jun;CHEN Ying(School of Computer Science and Information Engineering,Shanghai Institute of Technology,Shanghai 201418,China)

机构地区：[1]上海应用技术大学计算机科学与信息工程学院,上海201418

出　　处：《计算机工程与设计》2023年第6期1721-1728,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(61976140)。

摘　　要：针对协同过滤推荐算法中计算用户间相似度精准度低的问题,提出一种融合兴趣差异和评分差异的协同过滤推荐算法。结合标签信息和项目评分挖掘用户兴趣,利用信息熵度量用户间兴趣差异;为消除多种因素对项目评分数值作用效果的差距,以用户评分的最大差值为基准度量用户间评分差异;分析兴趣差异和评分差异定义用户间综合相似度并进行推荐。实验结果表明,该算法的用户间相似度计算结果比其它推荐算法准确。Aiming at the problem of low accuracy in calculating similarity between users in collaborative filtering recommendation algorithm,a collaborative filtering recommendation algorithm combining interest difference and rating difference was presented.The users’interest was mined by combining tag information and item-rating data,and based on that,the interest difference between users was calculated using information entropy.To eliminate the difference between the effects of multiple factors on the item-rating data,the maximum difference value of users’ratings was taken as the benchmark to measure the rating difference between users.By analyzing interest difference and rating difference between users,the comprehensive similarity between users was defined,so as to make recommendations.Experimental results show that the similarity between users calculated using the proposed algorithm is more accurate than that using other recommended algorithms.

关 键 词：个性化推荐 协同过滤算法 标签信息 兴趣差异 评分差异 信息熵 用户间相似度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]