基于项目特征与用户兴趣模糊性的推荐算法  

The Recommendation Algorithm Based on the Fuzziness of Project Features and User Interest

作　　者：黄向春 赵芬霞[1] 安建业[1] HUANG Xiang-chun;ZHAO Fen-xia;AN Jian-ye(School of Science,Tianjin University of Commerce,Tianjin 300134,China)

机构地区：[1]天津商业大学理学院,天津300134

出　　处：《软件导刊》2022年第9期14-18,共5页Software Guide

基　　金：国家社会科学基金青年项目(20CTJ011)。

摘　　要：为解决项目评分矩阵稀疏性问题,提出结合项目特征和用户评分的模糊性特性。首先利用类高斯模糊数描述项目所属类别的隶属度,然后应用梯形模糊数表示用户对项目的喜爱程度构建用户—项目类别偏好矩阵,最后构建基于项目特征的模糊性和用户兴趣的方法计算推荐信任分。在MovieLens 100k数据集上的实验结果表明,当N取1-300时,在不损失top-N推荐召回率的情况下,该算法相较于基于用户和基于项目的协同过滤推荐算法,平均推荐准确率分别提高了39.97%和5.74%,有效解决了数据稀疏性问题,可在历史行为数据较少的情况下,推荐用户感兴趣的项目。In order to solve the problem of sparse item scoring matrix,a fuzzy feature combining item features and user ratings is proposed.First,Gaussian like fuzzy numbers are used to describe the membership degree of the category to which the project belongs,then trapezoidal fuzzy numbers are used to represent the user′s preference for the project,and a user project category preference matrix is constructed. Finally,a method based on the fuzziness of project characteristics and user interest is constructed to calculate the recommendation score. The experimental results on movielens 100k dataset show that when n is 1~300,the average recommendation accuracy of this algorithm is improved by39.97% and 5.74% respectively compared with the user based and item based collaborative filtering recommendation algorithms without losing the top-N recommendation recall rate. It effectively solves the problem of data sparsity,and can recommend items of interest to users in the case of less historical behavior data.

关 键 词：协同过滤推荐算法 模糊数 用户兴趣模型 推荐信任分 用户偏好矩阵 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]