Information Confidentiality Using Fuzzy Based Data Transformation Method

The freedom and transparency of information flow on the Internet has heightened concerns of privacy. Given a set of data items, clustering algorithms group similar items together Knowledge extraction process poses certain problems like accessing sensitive, personal or business information. Privacy invasion occurs owing to the abuse of personal information. Hence privacy issues are challenging concern of the data miners. Privacy preservation is a complex task as it ensures the privacy of individuals without losing the accuracy of data mining results. In this paper, fuzzy based data transformation methods are proposed for privacy preserving clustering in centralized database environment. In case one, a fuzzy data transformation method is proposed and various experiments are conducted by varying the fuzzy membership functions such as Z-shaped fuzzy membership function, Triangular fuzzy membership function, Gaussian fuzzy membership function to transform the original dataset. In case two, a hybrid method is proposed as a combination of fuzzy data transformation approach specified in case one and Random Rotation Perturbation (RRP). The experimental results proved that the proposed hybrid method yields good results for all the member functions which are used in case one.