A new keyword-based similarity measuring model on peer groups in peer-to-peer networks

Peers were divided into different peer groups according to similar contents or interests to improve search efficiency in peer-to-peer networks. But with the number of peer groups grows up, when none peer could provide results for a request within one peer group, flooding will be used which will also cause large network flows. In order to increase search efficiency further, peer groups were studied, and the concept of similar peer groups and a keywords based similarity measuring model between peer groups were proposed. In the model, similarity degrees of each document belonging two peers were first calculated according to the keywords in different peer groups, and then the similarity degree between the two peers was determined by similar documents and their similarity degrees. After that, the two peer groups' similarity degree relied on the amount of the similar pairs and their similarity degrees. At last, simulation analysis demonstrated that the proposed model got closer results with the actual situation than models based VSM.