面向个性化推荐领域的超图表示及应用  

Hypergraph Representation and Application for Personalized Recommendation

作　　者：秦攀攀 潘文林[1] 张天军[1] QIN Panpan;PAN Wenlin;ZHANG Tianjun(School of Mathematics and Computer Science,Yunnan Minzu University,Kunming 650504)

机构地区：[1]云南民族大学数学与计算机科学学院,昆明650504

出　　处：《计算机与数字工程》2024年第12期3675-3680,共6页Computer & Digital Engineering

摘　　要：二分图刻画现实复杂系统中多元、多准则、多属性的关系时略显复杂。因此,论文提出用同构超图表示推荐系统中用户和项目的关系,用图形表示,集合表示和矩阵表示三种表示形式构建推荐系统超图模型,并根据超图的集合表示形式给出了同构超图的节点相似度的定义。从图形表示形式上看,相比较于用二分图描述推荐系统,超图的表示形式能够更形象、更直观地反映出用户和项目的交互关系。论文利用MovieLens数据集做了相应实验,将基于超图和二分图的推荐结果进行对比分析,结果验证了基于超图结构的推荐具备一定有效性,且提供了更好的推荐质量。The bipartite graph is slightly complicated when it depicts the relationship between multiple,muliple criteria and multiple attributes in the real complex system.Therefore,this paper proposes to use an isomorphic hypergraph to represent the rela-tionship between users and items in a recommender system.It uses three representations of graph representation,set representation and matrix representation to construct a hypergraph model of the recommender system.And the definition of node similarity for iso-morphic hypergraphs is given according to the set representation of the hypergraph.From the perspective of graphical representa-tion,the hypergraph representation can reflect the interaction between users and items more vividly and intuitively compared with the bipartite graph to describe the recommender system.This paper uses the data set of MovieLens to do coresponding experi-ments.According to compare and analyze the recommendation results based on hypergraph and bipartite graph,the result verifies that the recommendation based on the hypergraph structure is effective and it provides better recommendation quality.

关 键 词：同构超图 节点相似度 个性化推荐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]