Research On Prediction And Isolation Of Sensitive Information In Cloud Computing Environment

The cloud stores massive amounts of data,which are indispensable with sensitive,high-value sensitive information.If these data are not effective means to identify and protect them,it is vulnerable to attack and leakage.So how to identify sensitive information of different categories of intellectual bodies,and how to ensure their security is an urgent problem to be solved.This thesis proposes a sensitive information forecasting method based on fuzzy decision tree to find sensitive information in massive data and virtual isolation technology based on Chinese wall to realize the protection of sensitive information.The main research contents are:1.Sensitive Information Forecasting Method Based on Fuzzy Decision Tree.This thesis,the fuzzy decision tree algorithm is introduced into the identification of sensitive information.For the fuzzy of continuous attributes,the FCM algorithm is used to cluster the training set in the split nodes,and several classification intervals are formed,which are fuzzy in the process of building.For the choice of split attributes,the information gain is used as the standard,and the attribute with the largest value is selected as the next split attribute.The scale control of the decision tree is mainly achieved by stopping the condition and pruning algorithm.Finally,we construct the successful fuzzy decision tree to extract the forecasting rules,and then realize the prediction of sensitive information.This thesis,the simulation results are compared with other decision tree algorithms,and the error rate of the algorithm is relatively low.However,because of the clustering algorithm,the number of subnodes is relatively large.2.Virtual isolation technology based on Chinese wall model.This thesis uses the dynamic diversity of the Chinese wall model to achieve isolation protection of sensitive information in cloud computing environment.In the cloud computing environment,due to the highly virtualized resources,will make the identity of the host and guest for the change.Therefore,the thesis introduces the concept of diversity to blur the identity of the subject and object,and uses diversity to contain all the historical access information.First,there is a conflict between the specified diversity.Second,the access process of read and write rules are explained.Diversity can dynamically expand virtual resources according to read and write rules.Finally,the comparison of the isolation mechanism was carried out by experiment.The experimental results show that the sensitive information can not be protected before the isolation mechanism is added,and the sensitive information can be accessed with each other.After the isolation mechanism is added,the sensitive information can not be accessed from each other.In addition,the increase in the isolation mechanism,the diversity of the number of changes were tracked,the experimental results also meet the expected forecast results.In the cloud computing environment,the identification and protection of sensitive information has become a hot topic of current research.The main contributions of this thesis are as follows:(1)The fuzzy decision tree can be constructed in the existing sample space,and the forecasting rules can be obtained,and the prediction of sensitive information can be realized.(2)on the basis of the traditional Chinese wall,fuzzy subject and object,the relationship between the body,you can dynamically expand the access domain,to achieve the isolation of sensitive information protection.