Research Of The Web Site Credibility Evaluation Model Based On Cloud Computing

Due to the openness of the Internet,the Internet is filled with a lot of bad information,such as pornography and violence information,especially under the circumstance that the number of teenagers users in mobile device increases rapidly, it is necessary to evaluate the mobile terminal site. However,most existing web site evaluation models have shortcomings in the following two aspects: 1)unable to choose the evaluation factor set correctly,the evaluation factor set is repeated or is short of pertinence;2)ignore that website trust value changes over time.In view of the information security problem in mobile network and the shortcomings in the existing evaluation models,combining the cloud and the mobile terminal,this paper presents a web site credibility evaluation model based on cloud computing. Model is mainly divided into two parts,site classification and fuzzy comprehensive evaluation: first,classify the websites and choose the appropriate evaluation factor set and weight depending on the type of website,so that the model can eliminate the repeated content in evaluation factor set as well as assign weights reasonably.The evaluation results will be more reliability. This paper uses the multiple support vector machine(SVM) classification algorithm based on fast polynomial kernel function which means avoid a large number of support vectors participating in the calculation.So this algorithm greatly improve the efficiency of classification. Next is to evaluate the website according to the results of classification by using fuzzy comprehensive evaluation method.Because the website trust value changes over time as well as some sites may use unfair means to get trust value soared in the near future,the website trust value is divided into historical trust value and the current trust value in this model.Historical trust value reflects the credibility of the website in the past long time and the current trust value reflects performance of website recently,and also reflects the user preferences, habits and other information in the near period of time. The experiment shows that model is more reliable and personalized.In this paper, a web inspection system prototype is also implemented on the android platform,the system combines the cloud and terminal well.When user submits site inspection request in the terminal, the cloud calculates using the above model and algorithm and provides test result to the user. The system improves the terminal security on the Internet.