Computing Methods Research Of Recommendation Trust Value Between Nodes In P2P Network

As a new distributed network, The P2P network has broken the traditional internet computing model. The characteristics of no-central, scalability and robustness of P2P network make the communication between nodes great convenience, so it can be widely promoted, but at the same time, because of its openness, dynamic, autonomy and anonymity, the trust between nodes has seriously defected, and the trade between nodes also faces mevitable security problem. How to effectively and reasonably respond to the security problem is the focus of research in recent years. With the concept of trust management proposed, trust management model has became an effective means which analysis the trust relationship between nodes, compute trust value and ensure the information security of network.In P2P network, the calculation values of trust management model mainly include the direct trust value and the recommendation trust value. The existing trust management model has many more perfect methods in the calculation of the direct trust value between nodes, the algorithms are relatively mature, the calculation results are relatively accurate and reasonable, but in the calculation of recommendation trust value, especially the multilayer recommendation trust value such as recommendation chain and recommendation network, the methods still have many deficiencies, these problems that trust attenuation on the recommendation chain, trust synthesis of recommend chain, applicability of recommendation credibility and prevention against malicious recommendation have not been solved. This paper analyzes the relationship between trust and recommendation chain and combine type with characteristics of recommendation chain, proposes different computing methods and prevention mechanism against malicious recommendation according to the types of recommendation chain and networks in order to ensure the safety of transactions between nodes. The main work of this paper is as follows:Firstly, we put forward the concept of main chain and vice chain of recommendation chain, according to the recommendation link layers will recommend network recommendation chain for classify the recommendation network into simple and complex types according to the features recommendation chains; secondly, propose the concept of recommendation contribution degree, make the transaction history and recommendation history as the important factors of recommendation trust value computation; thirdly, according to the different types of recommendation network, put forward different recommendation trust value computing methods, a iterative algorithm is used in the rust value calculation of simple recommendation network and a suspected factor is added to against malicious recommendation, in the recommendation trust value calculation of complex network, an ideas of minimum recommendation trust is introduced, and use the algorithm of graph theory combined with hierarchical weighted directed graph to calculate the recommendation credibility, simulation experiments show that the algorithm is reasonable and effectively prevent malicious recommendation; finally, this paper analysis the trust value change caused by the recommendation effect after the transaction between nodes, and gives trust value and recommendation contribution updating methods.