A method of user anonymity in personalized retrieval oriented to large data

A method of user anonymity in personalized retrieval oriented to large data

Kanghaiyan Xiong Li

In order to solve the contradiction between the potential privacy security of personalized retrieval technology in large data and enhance the performance of personalized information retrieval, a method of user interest model anonymity based on the combination of difference privacy and P-link technology is proposed. First, the user's quasi-identifier generalization and add noise to meet the difference privacy protection requirements, maximize the query accuracy in the statistical database, while minimizing the probability of identifying individuals and attributes; Secondly, according to the similarity between users ' interests, it is p-link to satisfy the equivalence group, The weights of equivalent group interest items and the center of mass of equivalence group are calculated. Finally, the data that is anonymous is published. A large number of experiments show that this method combines the characteristics of difference privacy and p-link to realize the anonymity of the user's interest model and the user's interest is basically unchanged, which can protect the user's privacy information and ensure the personalized retrieval performance.

A method of user anonymity in personalized retrieval oriented to large data