Hypergraph Representation and Application for Personalized Recommendation
The bipartite graph is slightly complicated when it depicts the relationship between multiple,multiple 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 corresponding 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.
国家科技期刊平台
NSTL
万方数据
