Information Heredity Trust Model Based On Influence In P2P Networks

P2P networks have been becoming one of the most important applications in the Internet nowadays. Due to the scalability and flexibility, they are widely used in content sharing, distributed computing, and collaborative work etc. However, P2P networks are constituted by large number free users. They can freely join and leave the networks making the unstable quality of service. It results in a lack of trust among users. The absence of trust blocks the progress of the P2P networks badly.Trust is a comprehensive evaluation of the behavior and the ability of a user to another user. Trust among uses is the basis of the exchanges and cooperation, and the premise of cooperating and sharing. It directly impact on the P2P networks. By introducing trust mechanisms, trust model is one of the best way to solve the problem of trust in the P2P network. Learning from the knowledge of psychology, sociology and other disciplines on trust, we go into the trust models in P2P file-sharing application under a deep understanding of the nature of trust, and sorting out the relationship of the trust properties among them. The main achievements are as follow:The process of trust calculating based on influence is proposed. Learning from the process formation of the people’s view, the point of view to somebody (such as:trust or distrust) is formed based on own attitude with the influence of other people’s attitude. In the same way, the evaluation of trust is formed on the basis of the direct trust of trustor with the influence of recommended trust of others. In this dissertation, trust calculating based on influence is realized by introducing the concept of insisting degrees to quantify the influence and defining influence function.The selecting recommender mechanism based on interest. Learning from the interpersonal relationship in human society, attitudes of those who share the same interests are most valuable when evaluating a person. Through file’s DNA by learning from the principle of genetics, acquiring the list of providers and disseminators, my interest groups are gradually established. The members of interest groups will serve as candidates. The efficiency of selecting recommender is improved by querying the members of interest groups.Peers don’t malicious evaluate as much as possible because of themselves benefit by proposing the mechanism of linkage between evaluation and malicious peers, and evaluation feedback mechanism. On the one hand, according to the poor evaluation of a peer receiving the service, the system will put the service provider into a malicious list and avoid contacting with malicious peers. A peer would gradually isolate with normal service providers if it malicious evaluates normal service providers. On the other hand, a peer accepting service should feed back evaluation to the service provider. Providers will refuse to provide service or reduce the quality of service to malicious valuators so that the interests of malicious valuators will be damaged. Occurrence of malicious evaluation be inhibited possibly.Preliminary judgment of the authenticity of a file. Provider and disseminators are learned by the file’s DNA. According to the principle that the quality of a file is relatively more reliable with more disseminators and the list of malicious group, a preliminary judgment of the file authenticity is provided. Adaptive algorithm is given for the possible attacks of malicious Group.Reciprocal incentive based on interest groups. Interest groups are divided into a set of cooperation and non-cooperation by an autonomous peer itself. Under the guidance of game theory, different services are provided for corresponding sets with the spirit of reciprocal, punishment and tolerance. By gaming among peers, cooperation is promoted, and members of sets are dynamically adjusted according to its behavior.