基于标签属性的物质扩散推荐算法  

Material Diffusion Recommendation Algorithm Based on Tag Attributes

作　　者：卢兴才 孙更新 宾晟 LU Xing-cai;SUN Geng-xin;BIN Sheng(College of Computer Science and Technology,Qingdao University,Qingdao 266071,China)

机构地区：[1]青岛大学计算机科学技术学院,青岛266071

出　　处：《青岛大学学报（自然科学版）》2024年第3期32-39,共8页Journal of Qingdao University(Natural Science Edition)

基　　金：教育部人文社会科学研究青年项目(批准号:15YJC860001)资助;山东省自然科学基金面上项目(批准号:ZR2017MG011)资助;山东省社会科学规划项目(批准号:17CHLJ16)资助。

摘　　要：针对推荐系统中数据稀疏问题,利用用户与产品之间的标签信息,提出了一种基于标签属性的物质扩散推荐算法。通过构建用户—产品标签和产品—用户标签二部图网络,分别计算产品对用户标签属性和用户对不同类型产品的推荐偏好特征向量,据此改进物质扩散算法中的资源分配方式。在MovieLens电影评分数据集上的对比实验结果表明,该算法在精准率、召回率、F1系数和多样性上都有一定程度的提高。In order to solve the problem of data sparsity in recommendation system,a material diffusion recommendation algorithm based on tag attributes was proposed by using tag information between users and products.By constructing a user product tag bipartite graph network and a product user tag bipartite graph network,the recommended preference feature vectors for user tag attributes and for different types of products were calculated,and the resource allocation method in the material diffusion algorithm was improved accordingly.Comparative experimental results on the MovieLens movie rating data set show that the algorithm has improved to a certain extent in precision,recall,F1 coefficient and diversity.

关 键 词：推荐算法 二部图 物质扩散 标签推荐 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]