Research On Reputation Collusion Attack Recognition Algorithm Based On D-S Evidence Theory

With the rapid development of Internet information technology,e-commerce industry,especially online shopping is a high growth trend.While the brush single business platform selling fake,false promotions,price fraud and other fraud against the network integrity management came into being.As a new type of attack on reputation system,reputation collusion highlights the lack of trust management in E-commerce.Under the condition of lack of integrity in the whole economic environment,the lack of trust management restricts the healthy and rapid development of e-commerce.In this paper,we focus on the research of ecommerce reputation collusion attack identification method,one can reduce the risk of online transactions,improve consumer online shopping behavior,improve the online reputation system,create a good environment of electronic commerce credit,on the other hand can be an effective complement to traditional security mechanisms,to achieve a higher level of security policy,to combat cyber fraud to provide theoretical and methodological support for the assist.In this paper,we focus on the problem of e-commerce reputation collusion attack,and use the D-S uncertainty reasoning theory,in order to improve the collusion attack recognition effect as the goal,mainly study measure of evidence conflict,identification algorithm based on weak decision fusion,and identification algorithm based on evidence reliability measure,specific contents are as follows:1.Aiming at the problem that the existing methods of measuring the conflict of evidence can not accurately represent the decision difference of the two evidences in reputation collusion,this paper takes into account various factors such as potential decision making,mass value and structural relationship and presents a method to measure the conflict of evidence based on the decision making distance.The experimental results show that the presented method can be used to describe the difference of evidence independent decision,and measure the conflict between evidences more reasonably,and it is an important foundation for the research of the following recognition algorithms.2.Aiming at the problem that the reputation collusion attack has strong simulation,the extracted features have a weak ability to distinguish attack or not,then the quantitative evidence is weak decision-making evidence,this paper proposes an algorithm based on weak decision-making evidence fusion.The proposed recognition algorithm according different situations the influence of weak decision-making evidence for evidence fusion,give full play to the decision-making function of weak decision-making evidence,reduce the uncertainty decision of fusion results.Multiple performance indicators for multiple experiments show that the proposed algorithm not only can effectively identify two types of reputation attack attacks,but also can be extended to multi class classification problems,has a good expansibility.3.Aiming at the problem that diversity of evidence sources in reputation collusion attack,the reliability of evidence is different,direct use classical D-S synthesis may lead to results deviate from the true situation,this paper proposes an algorithm based on evidence reliability.The proposed recognition algorithms for different situations design the two-way multiple evidence reliability measurement,can accurately measure evidence reliability,and improve the accuracy of the recognition algorithm.Multiple performance indicators for multiple experiments show that the proposed algorithm not only can effectively identify two types of reputation attack attacks,but also can be extended to multi class classification problems,has a good expansibility.