Research On Open Network Oriented Subjective Trust Model

At present,with the development of the open network,the network environment presents the openness,sharing and dynamic characteristics.Open Internet environment provides much convenience for people's life online,such as network shopping,instant messaging,BBS,etc.The rapid development of Internet in convenient people's life style also brought many hidden dangers,such as false advertising,online ghostwriters,malicious comments,etc.These questions are the key and difficult point for the current trust research.Currently,according to the mathematical tools used by the difference in the field of trust,subjective trust model mainly divided into two categories:One kind is a trust model based on probability theory,the other is a trust model based on fuzzy theory.At present,most trust models are the trust model based on probability theory,but the precise mathematical model can not reflect the trust of the subjectivity and fuzziness.In addition,these models are only static analysis of the trust,and neglected the dynamic.To solve above problems,this paper puts forward three kinds of subjective trust fusion models,the specific content as follows:First,in order to overcome the problem of lack dynamic in the existing trust model,this scheme adds human cognitive factors into trust research.Considering interaction as the influencing factors of cognitive process,according to the fuzzy comprehensive evaluation algorithm to calculate cognitive factor,combined with the direct trust and indirect trust and reputation building trust fusion model.This trust model combining with human cognitive processes,the trust set up process associated with the process of cognition,which good embodies the dynamics of trust.Second,aimed to the fuzzy of trust,and consider trust relationship among various factors,this paper proposes a new trust model.This model combines the Naive Bayesian algorithm and fuzzy theory.It makes direct trust,indirect trust and reputation as trust characteristic attributes.According to Naive Bayesian theory,the prior condition probability is used to calculate the posterior probability of trust interactive results.This model will help truster to make decision analysis by the result of interaction.This model overcomes the problem of fuzziness and uncertainty insufficient in probability theory trust model.Third,due to the subjectivity and uncertainty of trust,the precise mathematical tools such as probability theory cannot describe the essence of trust.On the other hand,the fuzzy theory is very suitable for dealing with fuzzy and uncertain information.however,in existing trust model,the fuzzy theory are based on individual preferences or experience value to distribute weight,and does not take into account the characteristics of trust changes over time.This paper analyzes the factors of trust in different periods to build the judgment matrix by analytic hierarchy process.Then,distribute weight for each factor.At the same time,aim to the malicious attack in open network,this model with Lattice close-degree punishes malicious recommendation and rewards kindness recommendation to reduce the risk of interaction.