Content-Based Community Formation in Hybrid Peer-to-Peer Networks.
Atip Asvanund, Ramayya Krishnan · 2004
We present a community formation method for improving IR performance in a distributed digital library network. Following the digital library literature, we model digital libraries as leaf nodes and regional directory services as ultrapeers in the hybrid P2P architectures. We conceptualize a regional directory service and its local network of digital libraries as a community, and introduce a community formation method in which digital libraries prefer to connect to the best regional directory service. Conversely, regional directory services prefer to recruit the best digital libraries to its community, and to establish connections to the best adjacent regional directory services. Our method utilizes content-based techniques to identify users ’ content interests and IR patterns in order to evolve the network towards a topology that improves IR efficiency. �e method operates in a decentralized and transient setting without the assumption of a central social planner and uniquely identified content, which are typical assumptions in other work. Finally, we take precautions into examining measures for reducing the overheads in applying our method and demonstrate conditions in which the benefits in applying our method outweigh its costs. �is work introduces an efficient solution that is readily implementable in the practical assumptions of today’s P2P networks. 1