A Review Model For Protecting Location Privacy Based On Negative Surveys

In recent years,Dazhong Dianping,Yelp,and other online reviews systems have been widely used to help people make better consumption decisions in their daily lives,and have received high attention.However,they list all comments submitted voluntarily by users.The information attached to the comments poses a threat to the user's location privacy.Traditional location-privacy protection methods mainly convert location information through generalization,perturbation,and so on.In these systems,location information attached to the comments needs to be presented accurately to maintain the usability of comment information.Therefore,traditional location-privacy protection method is not applicable.This thesis proposes a novel review model based on negative survey to protect location privacy.Negative survey is inspired by the negative selection mechanism in Biological Immune System,which can collect user data and estimated the aggregated results while protecting privacy.Different from the existing methods,the method of this thesis disturbs the real user information(ID)of the comments to protect the user's location privacy.Additionally,there is usually some background knowledge in realworld applications,which can be used to improve the accuracy of aggregated results.A reconstruction algorithm for background knowledge is also proposed in the thesis,which can handle the background knowledge more effectively than conventional reconstruction algorithm.The specific work is as follows:1)This thesis proposes a review model based on negative survey to protect location privacy.Firstly,the model analyses the active areas and preferences of users based on the location information in the comments.Then,users are grouped according to their active areas and preferences.In a group,applying the negative survey for users' comments can disturb the user ID of each comment,which makes the attacker unable to identify the real publisher of the comments.Finally,the reconstruction algorithm of negative survey is used to estimate the number of comments for each user in a group.Such feature guarantees the availability of functions related to the number of user comments in the system,such as calculating the user's level in the system.This thesis collects the real comment data of users in Dazhong Dianping,and verifies the validity of the model.2)When there is background knowledge,this thesis proposes a review model based on background knowledge and negative survey to protect location privacy.A negative survey reconstruction algorithm NStoPS-UD is described in this work,which can effectively make use of background knowledge.We use this algorithm to protect location privacy effectively in our review model.In the experiment,simulation data and the data of Dazhong Dianping is carried out.Intensive experiments compare the proposed algorithm with the existing methods.Results show that the proposed algorithm can handle both negative values and values that do not satisfy the background knowledge,and the error is smaller.Finally,the NStoPS-UD is extended to the negative survey when the selection probability satisfies arbitrary distribution,and the formula is given.