A Proposal of Access-based Contents Recommendation in Virtual Database Environment on P2P Network

摘要

In recent years, EC sites offer various and vast amount of products. In most of the large EC sites, they provide recommendation service of the products for customer's convenience and for the promotion of the sales. However, in such a recommendation service, since only the contents which the EC site has managed are the objects of the recommendation, recommendation of contents from more than two EC sites combined are difficult to perform. In this study, we build a virtual retrieval system with ad hoc participation of databases in P2P network environment. It becomes possible to access one huge virtual database by integrating databases which participate to the P2P network. In addition, we propose a contents recommendation service which spans multiple databases by performing clustering of contents of multiple databases based on the contents access history and frequency. By constructing an experimental contents retrieval system, we evaluate the efficiency of the proposed clustering method and contents recommendation mechanism.