Homepage Finding in Hybrid Peer-to-Peer Networks

This paper illustrates a ranking scheme which combines fulltext, anchor text and URL structure for homepage finding in hybrid peer-to-peer networks. The experimental results show that the proposed ranking scheme permits to find the correct homepages after exploring a small portion of a hybrid, hierarchical peer-to-peer network when the group of peers which includes the peer that stores the correct homepage is selected. A large, highly heterogeneous Web data test collection searched by homepage finding queries was used. The low computational overhead and the good effectiveness allow for an implementation in information retrieval environments.