Gossip Strategies for Service Composition

摘要: Unstructured peer-to-peer (P2P) architectures offer several benefits to implement semantic discovery and composition in future-generation service registries. However, their success strongly depends on the adoption of efficient techniques for disseminating queries over network. Gossip strategies significantly reduce amount messages with respect flooding, but they need a predefined tuning effectual fanout achieve good performance. In this paper, we compare typical gossip our proposal, which is able dynamically exploit network knowledge fulfil selective choice propagation paths order ensure high recall further number exchanged. We perform comparison simulated environment observe resolution time, message overhead large-size evolving networks while searching compositions. have adopted Bernoulli, Random Geometric Scale-Free graphs model different topologies. The experimental results show that approach adapt changes preserve levels recall. particular, it reduces overhead, both optimized flooding analysed gossip-based strategies, or improves recall, whereas time remains almost unchanged.