A Level K-Anonymous Technology Of Graphlet Structure Perception On Social Network Publishing

In recent years,social networks have become more and more popular,and many large social platforms have emerged,such as Momo,WeChat,Renren,etc.Now most people use these social platforms to communicate with family,friends and colleagues.This phenomenon has also caused many people to share some personal information on social networking platforms,therefore,social networks has become an important source of research and data mining in many areas.However,this direct data embeds many users' private information,even sensitive information such as identity,salary,etc.Therefore,if the user's original information is published directly,the user's personal privacy will be revealed.Therefore,before the release of this data information,privacy processing is required.How to deal with it in order to balance protection and utility is a problem that has been researched.In the privacy protection methods of existing social networks,the low-order connected organizational structure(nodes and edges)or the uncertain structure(such as the community structure)is always considered,and the high-order connected organizational structure(graphlet/motif)is not considered.In reality world,the high-order connected organizational structure in the complex social network occupies an increasingly important role in the analysis and research of social data.Therefore,how to preserve the graphlet structure information in the social network graph has become a key issue in privacy protection.Higher-order connected organizational structure has many applications in social network analysis.For example,in the transportation network,the graphlet will be regarded as a basic unit to measure the accessibility of traffic.Therefore,protecting the graphlet structure will greatly improve data utility.The main work of this paper is as follows:First of all,there are many kinds of current privacy protection methods,including k-anonymous,randomized and differential privacy,etc.After analyzing,we find that it is insufficient.Firstly,when randomization and differential privacy randomly disturb the graph data,it will produce more damage to the structures.The traditional k-anonymity method considers only low-order connected organizational structures or uncertain structures in the process of graph processing.The second: The social network graph obeys power-law distribution,the traditional k-anonymity method does not take this property into consideration.It adopts a unified privacy policy for all node degrees.This will not only damage the graphlet structure in the social network,but also too much disturbance to some nodes in the graph,thereby reduce the value of social network data.Secondly,for the shortage of the above methods and the application of the graphlet itself,this paper proposes a level k-anonymous technology of graphlet structure perception on social network publishing.This method considers the degree of social network node according to the characteristics of power-law distribution,the nodes are divided according to the degree,and the analysis graphlet structural features of the graph in the privacy process and adjusts the privacy-processing strategies of the edges according to graphlet structural features,thereby in order to meet the privacy requirement while protecting graph structure information in the social network,improve the utility of the data.Finally,this paper uses two real public data sets,WebKB and Cora,conducted experiments and evaluation.Finally,the experimental results show that the method proposed in this paper can provide the same privacy protection intensity,at the same time,it can better maintain the social network's structural information and improve the data's utility.