An Efficient Framework for Privacy Preserving in Correlated Data based on Tree Distribution Approach

Privacy preserving in Data Mining is study of achieving some data mining goals without scarifying the privacy of individuals. In this paper privacy has been provided to correlated dataset in which users fires a query to get the output. In general attributes in a dataset are sampled independently. However, in real-world attributes in a dataset are rarely independent. If one attribute is depend on another then it leads to privacy violation. The solution to this problem is provide privacy using correlated tree distribution approach in which query gets scrutinized to get proper attributes in the dataset. Different steps are processed like K-Means Clustering algorithm, Correlation Computation, Tree Creation. If-Then Rules are used to set ranges where the sensitive information has been resided. Access control policies are provided in order to protect information. Different mechanisms are used, like relaxed admissible mechanism Correlated Iteration Mechanism (CIM), Privacy preservation approach via data set complementation to maintain privacy. Many existed system have a drawback in which major drawback is all have huge time complexity. In tree distribution approach the drawback has been overcome in which it strictly focuses on finding correlation amongst data. To enhance the privacy in Non-IID (Independent and identically distributed) datasets proposed system uses tree based traversing technique for maintaining a strict access control over the attributes. The main aim of the system is to provide privacy when user searches globally.