Study On Key Techniques Of Query Processing Over P2P Network

P2P (peer-to-peer) network breaks the traditional client/server mode, improves the network efficiency, makes the best of network bandwith, and develops the potential of network nodes. P2P network is featured with high scalabilty and fault tolerance. So, P2P has become the focus of both computer academic and industrial sectors. Query processing is the key problem of P2P network, and due to the distribution, dynamic and heterogeneity of P2P network, there are some new challenges and opportunities for query processing over P2P network. The approaches about routing and location, comlex query processing, multi-keyword query routing in P2P Web search, information retrieval in social network, resource search in unstructed P2P network are researched in this dissertation. Here our work includes the following major aspects.(1) A novel structured routing protocol with semantic locality and high fault tolerance called RingS is presented. RingS provides support for key-based lookup service, and implemtents effective data location. It can provide high routing efficiency of the logarithmic order, strong fault tolerance, explicit semantic locality and generality. In dynamic network environment, especially when network partitioning takes place, it could ensure most nodes’ functionality. The experiments are conducted to evaluate the performance and verify the advantages of RingS. The results show that RingS has good performance.(2) A Termset-based multi-keyword query processing strategy is presented. The correlations among keywords are taken into account, and the indexing and query processing are based on Termset. Moreover, several methods are applied to reduce bandwidth consumption. The network communication cost resulted by single keyword based indexing and distributed intersection operations during multi-keyword query processing is decreased greatley. The experiments show that our method has good performance.(3) An effective benefit based multi-keyword query processing strategy for efficient query routing is presented. Inspired by the observations on the distributions of query keywords and various features on the scenario of P2P Web search, we mine the correlations of keywords as well as coverage and overlap of the peers, and then we applie Nash Equilibrium to construct the optimal peer group based on benefit. Moreover, we utilize min-wise independent permutation for overlap detection to avoid futile routing of the same result, thus the network communication cost are reduced further. The experiments demonstrate the efficiency and effectiveness of our approach.(4) A query processing strategy supporting information retrieval in social network called PIST is presented. PIST is layered on top of RingS, a structructed P2P network. And PIST provides the distributed index and query processing mechinism. Further, user’s interests, friends, and feedback are exploited to provide personalized search results, and the performance of search is improved. The experiments demonstrate the feasibility and effectiveness of PIST.(5) A resource search strategy based on semantic clustering called SC-Search is presented. We propose an algorithm of constructing semantic forest to cluster the local resources of node. Then we establish similar links among nodes, so as to cluster the nodes with similar resources and utilize these similar links to process query. Further more, optimizing these similar links dynamically makes the nodes nearer in physical network. When forwarding the query, we combine the orientation mode and flooding mode, and add a dynamic TTL mechanism, so that the success rate and efficiency of resource search is improved further. Finally, the experimental results verify the feasibility of SC-Search.(6) We design and implement the prototype system of P2PSearcher, which applies the theories and approaches about routing protocol, complex query processing and information retrieval of social network proposed in this dissertation. The system shows validity and efficiency of these theories and approaches.In summary, this dissertation is dedicated to the fundamental problems related to query processing over P2P netwrk, including routing and location, complex query processing, information retrieval in social network and resource search in unstructured P2P network. Lots of theoretical analysis and experiments show that these approaches are efficient and effective. These approaches and techniques could make some contributions to the research work on this field.