Research On K-anonymity Method Based On Fuzzy Clustering

In recent years,the number of social networks has exploded in order to understand user behavior patterns and the main purpose behind data mining.Social software has begun to enter our daily life in a comprehensive way.At the same time,a large amount of user data has been published on the Internet.Proper use of these data can help society cope with problems better.However,while sharing these user data,there may be problems such as disclosure of user privacy data,which requires the intervention of privacy protection technology.At present,the most common privacy protection technology is to anonymize data by deleting or modifying some information,and the requirement of anonymization is to retain the information of the original data as much as possible.Most of the existing anonymous technologies cannot resist identity attack,attribute attack,link attack and similarity attack simultaneously.In addition,some of the existing anonymous technology in the process of anonymity will produce a lot of information loss and data distortion.In response to the above problems,a multi-constrained objective function privacy protection method based on K-member fuzzy clustering is proposed to protect anonymous data from identity leakage,attribute leakage,correlation leakage and similarity attacks,while minimizing data modification strategies,To reduce the loss of information.The method first proposes an improved algorithm for K-member fuzzy clustering.Then the multi-constraint objective function is used to further optimize the clustering center,while satisfying the Kanonymity,L-diversity and T-closure constraints,the clustering error rate and the loss of generated information are minimized.The experimental results of this paper show that the proposed method of privacy protection with multiple constraints objective function based on K-member fuzzy clustering can greatly reduce the information loss of data,and At the same time,it meets the constraints of k-anonymity,l-diversity and t-closeness neighbor.