Privacy Protection,Methods In Social Network Data Publishing

Social networks have become very popular in the past few years,because they allow users to express their personality and get to know people with similar interests.Despite this,there are many potential threats to the privacy associated with these users,such as identity theft and the disclosure of sensitive information.However,many users still unaware of these threats,and the privacy settings provided by social networks are not flexible enough to protect user data.There is no way for social network providers and governments to effectively protect users from privacy violations.But efforts need to be made to change this situation.In view of the privacy leakage in the current data publishing,this paper mainly undertook the following work:(1)In terms of current issue that clustering results in data clustering analysis of data publishing are susceptible to the initial center point,a differential privacy based MPDPk-medoids clustering method is proposed.This method is based on the traditional k-medoids clustering algorithm,through the optimization of the initial center point,and adding a small amount of noise in the process of clustering to improve the data availability and user privacy information.In addition,we propose a system model for social network user data publishing,which ensures that the user's private data is well protected at all stages of data publishing,ensuring user data security.By comparing with the existing privacy protection clustering methods,the experimental results show that the proposed MPDPk-medoids method is superior to the existing methods in terms of privacy protection and data availability.(2)Refering to the current issue of high data generalization and low privacy in traditional micro-aggregation algorithms,l-diversity based MDAV micro-aggregation privacy protection method(k,I)-MDAV is proposed.The method.improves the availability of data by reducing the degree of data generalization in the process of generalizing logarithmic data.Then,by using the micro-aggregation algorithm,the processed data set is divided into equivalence classes,and waiting for In the process of price class division,the degree of data privacy protection is improved by making it satisfy the 1-diversity.The specific experimental analysis shows that the proposed method can achieve a better balance between data validity and privacy protection.(3)Regarding present problem of low data availability and privacy in traditional equivalence class division,a privacy protection method based on equivalence class substitution is proposed.The method firstly realizes the minimum information loss in the same equivalence class by calculating the distance between each tuple in the record,and then publishes the equivalence class by selecting the equivalence class that is most similar to the release equivalence class.The replacement method implements privacy protection of user data.By comparing with the existing privacy protection methods based on equivalence classes,the experimental results show that the proposed method has a good performance in data availability and privacy protection.