Research On The Trust Evaluation Model Based On Multi-agent Under The 4PL Environment

With the deepening constantly of economic globalization and the growing prosperity of e-commerce, the development of the logistics tends to be multinational. As a result the concept of the fourth party logistics(4PL) is derived. In recently, Alibaba(Group) and Intime Retail(Group) has take part in China Smart Logistic Network(CSN) to build a 4PL platform. The 4PL platform improves the efficiency of logistics by integrated the third party logistics(3PL). Since the 4PL platform is running on a social network, the intention of the participants is unknown, thus the individual participate in the platform cannot trust each other and therefore they cannot have normal activities. However the traditional security method is hard to guarantee the safety of such a system. With the combination of virtual society and real one, and with the prediction and expression of system behavior, these problems can be solved by trust model based on multi-agent.Based on the analysis of the development and research status of trust model at home and abroad, this thesis proposed a trust evaluation model based on multi-agent under the 4PL environment and a kind of trust evaluation algorithm. First, we consider the agent`s autonomy in 4PL, we study the trust model of 4PL mainly base on multi-agent theory. Then, this thesis present a trust evaluation algorithm base on the individual difference(IDT). This algorithm improving the accuracy of the assessment by modify the feedback trust with calculate the individual difference. At last, A simple Multi-Agent system under 4PL environment has been realized base on the trust model and algorithm this thesis proposed. Finally, this system has been tested on functionality and performance. According to analysis of different trust model simulation test data, the trust model of this thesis proposed has good performance on improving accuracy rate and success rate.