加入用户和项目属性的奇异值分解推荐算法  

Singular Value Decomposition Recommendation Algorithm With User and Item Attributes

作　　者：李浩 潘莹 梁京章[1] LI Hao;PAN Ying;LIANG Jingzhang(School of Electrical Engineering,Guangxi University,Nanning 530004,China;Information Network Center,Guangxi University,Nanning 530004,China)

机构地区：[1]广西大学电气工程学院,广西南宁530004 [2]广西大学信息网络中心,广西南宁530004

出　　处：《装备制造技术》2021年第12期52-57,共6页Equipment Manufacturing Technology

基　　金：赛尔网络下一代互联网技术创新项目(NGII20190108)。

摘　　要：传统推荐算法仅利用用户的评分行为进行相似度计算来实现目标推荐,未考虑用户和项目的属性信息,推荐精度低,因此提出加入用户和项目属性的奇异值分解推荐算法。该算法在奇异值分解算法中加入用户和项目的属性信息,并在此基础上设计一种新的相似度计算方法,在相似度计算中为属性分配不同的权重,充分考虑每一种属性对推荐效果的影响,从而为目标用户和项目找到最相似用户集和最相似项目集。在数据集MovieLens上的实验结果表明,其在召回率、精确率值上均优于传统推荐算法。The traditional recommendation algorithm only uses the user's scoring behavior to calculate the similarity to achieve the target recommendation,does not consider the attribute information of users and projects,and the recommendation accuracy is low.Therefore,a singular value decomposition recommendation algorithm adding user and project attributes is proposed.The algorithm adds the attribute information of users and items to the singular value decomposition algorithm,and designs a new similarity calculation method.In the similarity calculation,different weights are assigned to the attributes,and the influence of each attribute on the recommendation effect is fully considered,so as to find the most similar user set and the most similar item set for the target users and items.The experimental results on the data set MovieLens show that it is better than the traditional recommendation algorithm in recall and accuracy.

关 键 词：推荐算法 奇异值分解 用户属性 项目属性 相似度计算 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]