Research And Implement On The Strategy Of Trace Hiding In Web Query Based On Group Signature

Internet service providers(ISPs)collect Web queries of users,then analyze the behaviors and intentions in order to improve the quality of service.With the development of "cross-screen marketing" model,ISPs obtain queries accurately by private account login.Search engines mine query intentions of visitors to return the satisfactoty results.Advertisers analyse the attributes of every user to push targeted advertisements.Hackers steal user privacy by hijacking query data,such as account and bank card passwords.Many users want to get good service but do not want to disclose private information.Some existing Web query privacy protection technologies such as anonymous network and "Track Me Not"(TMN)have following problems:(1)They can not protect privacy essentially,ISPs still identify the single user.(2)They can not play a protective role if searching by account login.To solve the problems and provide personalized service to users,we propose the strategy of trace hiding based on group signature in the thesis.In a group signature scheme,any member can sign the message on behalf of the group anonymously.In our thesis,any member searchs a query on behalf of the group.The scheme includes two parts:(1)group establishment.When a user apply to join the goup,Trust Center(TC)verifies the legitimacy firstly,then issues RSA private key and AES seed key to each group member.At last,TC sends RSA public key,AES key and identity informtion of every member to group manager.(2)HTTP proxy server/Group manager.Adding two functions based on general proxy server.Each member in the group sends requests to proxy server and proxy server automatically logs in website with public account instead of members'.And the response from proxy to a member is encrypted to ensure the confidentiality.In the experiments,the scheme applies RSA and AES encryption function and decryption function in the Openssl library.HTTP proxy server collects all member's Web queries,then we use WEKA which is a data mining platform to classify the Web query log.WEKA analyses query dataset according to query length,weekday/weekend and sensitivity.Comparative experiment shows that queries can be identified with high true positive rates.So TMN can not hide the user status and query information.Experiments in this thesis illustrate that our strategy can hide attributes of user's query in dataset,because query patterns from a group is a mixture of members and Web server can not analyze and mine a collection of a single member query attribute.Therefore a user's query trace on Web is proteceted.