Research On Adptive Dynamic Trust Measurement And Prediction Model Based On Behavior Monitoring

In recent years, with the Internet-based platform, a variety of large-scale distributed applications (such as grid computing, P2P(Peer-to-Peer) computing, electronic commerce, Ad hoc and pervasive computing, etc.) in-depth study, system performance, as there are several software collaborative model of dynamic service composition which model the dynamic nature of trust is one of the most complex concepts in social relations, involving assumptions, expectations, behavior, and environmental and other factors, it is difficult accurately quantified and predicted. Therefore, how to quickly and accurately for open distributed environment determine the credibility of an entity, multi-dimensional data model to overcome the traditional lack of processing capacity to become a problem to be solved, this is the basic issue for further research.First, review the emergence and development of trust management, and pointed out the dynamic relationship of trust modeling and management technology is a trust management technology in order to adapt to the development of Internet, the emergence of new directions. Describes the relationship of trust in distributed systems dynamic model, a careful analysis of the existing dynamic trust model as well as their advantages and disadvantages, are given some ideas to improve these models.Secondly, the trust from the classical mathematical theory of rough set model and get inspiration, proposed an improved adaptive behavior-based monitoring of dynamic trust model, describes the trust value for the definition and mathematical description, and then gives the behavior of data access method and with the process.After that, the confidence measure is given a new prediction method, based on rough set theory test build trust in knowledge representation systems, knowledge acquisition and classification categories is given the weight calculation algorithm for a complete description and theoretical analysis of algorithms.Finally, experiments using the campus network system and method of combining matlab program to simulate the experiment, the model accuracy and scalability of the dynamic aspects of the proposed trust model to demonstrate measurement and analysis.