Resource Search Algorithm Based On P2P Network

With the development of network technology, the number of users of the Internet becomesmore and more. Traditional centralized C/S network mode is facing more and more problems andchallenges due to the failure of single point and the large load of server. P2P network came intobeing as a new type of network model. The concept of central server is weakened in P2P networkmode, in which nodes can either act as a client and can act as a server-side. This mode can be agood solution to the lack of C/S network model, which is widely concerned by the academic andresearch community.P2P network model has been widely used in resource sharing, collaborative office,distributed computing, streaming media transmission and instant messaging, and other fieldsbecause of its good scalability and self-organizing of nodes. A core problem of these applicationsfacing is how to obtain the resources needed by the user in the vast amounts of dataresources,therefore, how to establish an efficient mechanism to locate resources has become ahot topic of research of P2P networks. Firstly the P2P network resource search mechanism isin-depth studied,and then unstructured peer-to-peer networks and structured peer-to-peernetwork resources search algorithm were in-depth analyzed.To unstructured peer-to-peernetwork, we put forward an improving search strategy and hope to decrease the redundantmessages of network and improve search efficiency and accuracy. Semantic retrieval strategy forstructured peer-to-peer network can only provide a single keyword based matching querywithout semantic query,the aim is to improve the query success rate and reduce network latency.The main contents of this paper are as follows:Firstly, the knowledge of P2P networks is analyzed in detail which Including research statusof home and abroad, the background and significance of the research, the concept, characteristics,application and topology of P2P network, in addition, the typical resource search mechanism ofP2P network is highlighted.Secondly, an resource search strategy based on top-k query of unstructured P2P network isproposed to improve the phenomenon of the large lode and the low query efficiency existing inFlooding algorithm. In this strategy,we use a vector space model VSM to describe the documentas the form of vector with weight, and we consider the probability of the nodes forwardingmechanism based on document similarity and the attraction of nodes when we select the nextforwarding nodes; Accoring to the forwarding formula,the neighbors that meeting forwardingconditions are selected to forward query. At the same time,the parter nodes,that has the greatestinteresting correlations with the local nodes are selected in this strategy. These parter nodes areupdated periodically, which can help to select the forwarding path; Taking into account that the users might not care about all the return results, and only care about a small part, a query tree isintroduced to return the top-k optimal results that meeting the conditions. At last,through thesimulation, no matter the load of node, and the precision of search, this algorithm will play wellcomparing with the flooding search algorithm.Finally, the P2P semantic retrieval strategy based on the structure of the body and VSMmodel is put forward. The strategy is the improvement of the Chord search model.In thealgorithms,we use the vector space model to describe the document to obtain similar angle, andthen map to space identifier vector in accordance with the principle of the DHT, so that theoriginal <key, value> pair of Chord model is improved as <vector, key,value>; After the resourceindex node,the ontology technology is used to return the semantically related documents to theusers matching on the semantic level semantics related documents. In addition, the nodemaintain the original routing table and a partner node table, storing the greatest interest relevantnodes to the local node, which can speed up the semantic retrieval speed. At last, we can see thatno matter the rate of search recall and reducing the delay of search,which is effective throughthe experimental results.