Research On Trust Management Model And Method In Electronic Commerce

How to establish, maintain, promot the bilateral trusting relationship, to reduces the electronic commerce transaction risk, to reduce the transaction cost, to guarantee the transaction completed safty and smoothly, have become an important question urgently to be solved in the electronic commerce development. The formal model of electronic commerce trust management, electronic commerce trust management model based on reputation and electronic commerce trust model based on the Bayes network have been researched in this paper aimming to the application.The major research efforts and contributions are as follows:(1) Classify the electronic commerce trust model from the multiple perspectives. The research is given from three perspectives:the trust research methods, the trust type, the type of trust value, particularly carried on the comprehensive classification according to the trust research methods, which gives good research thinking for further study.(2) Discuss the definition of trust in electronic commerce, and the properties of trust are given. Research the way to build trust relationship, meanwhile the initinal Trust model and dynamic trust model in electronic commerce are given.(3) Analyse the existing electronic commerce trust management formal model, point out its insufficiency, and propose formal model of trust management in electronic commerce (FMTM), which provides a theory frame for the development of trust management system.(4) Propose an electronic commerce trust management model based on reputation (ECTMBR). The model takes into account the influence of node trust from foll wing factors: the transaction amount, the trading hours, the transaction number of times, the transaction evaluation, and the risk factor and so on. The reputation in the model is composoed of the direct reputation and the indirect reputation, the direct reputation is caculated by above fives factors, and indirect reputation is calculated by friend recommendation and the stranger recommendation. The concept of belief degree is introduced, direct reputation belief degree is calculated by transaction amount, and indirect reputation belief degree is calculated using recommendation node quantity and the separate degree of recommendation information throgh S characteristic function and the standard deviation. Finally the computation formula of total reputation is given, which rejects the influence from human factors. Experiment results and analysis show that the model has a good validity and anti-malicious node's ability.(5) Propose an electronic commerce transaction trust model based on the Bayesian Network (TMBBN).The specific structure drawing of Bayesian network is given, and several main factors which have influence on transaction are discussed using the conditional probability formula. The time attenuation factor and the rewards, punishment mechanism are introduced in this model. The total reputation's computational method and its updation process are also given. It shows the experiment simulation environment, concrete algorithm and the simulation flow chart are given, finally the experiment data is analysed. The results show that this model can compute the reputation simplely and accurately, meanwhile it has certain anti-attack capability.