The Research On Recommender System Based On Substitute Goods Model

Information filtering which is derived from information retrieval is attracting more and more attention. Using information filtering technology in the results presented by information retrieval engine can greatly improve the retrieval accuracy. In the meantime, information filtering can offer some individual services which can't be gotten in information retrieval. Recommender system is the application of information filtering technology. There are two kinds of filtering technologies in traditional recommender systems. They are content based filtering and collaborative filtering.In this thesis, I present the definition of'related collection'on the based of traditional filtering architecture and'substitute product'in economics. When the increase number of product A scale with the decrease number of product B, we call A and B substitute products. When one product's changing number scale with another product's we put them in'related collection'. And the scale rate is called'related rate'.'Related collection'is the extension of substitute products. It includes positive relation and passive relation. When two products change in the same way...