pRoute: Peer selection using shared term similarity matrices

Abstract

Peer-to-Peer systems have proven to be an effective way of sharing data. The focus of this paper is on distributed search based on Peer-to-Peer technology. In this paper we present the pRoute system where peers advertise a short description of the content that they share, namely a set of terms. Peers remember the advertisements of related peers and thereby form a semantic overlay by which we mean that peers with similar content are grouped together. Peers calculate the similarity between their content descriptions by a term similarity function which, in the ideal case, is identical for all peers. In simulation experiments we compare the performance of different advertisement- and forwarding policies with respect to precision, recall and the number of messages. The results indicate precision and recall increase when the policies take semantics into account, without an increase of the number of advertisement- and query messages.