Data Security Using Decomposition

Protection of privacy from unauthorized access is one of the primary concerns in data use, from national security to business transactions. It brings out a new branch of data mining, known as Privacy Preserving Data Mining (PPDM). Privacy-Preserving is a major concern in the application of data mining techniques to datasets containing personal, sensitive, or confidential information. Data distortion is a critical component to preserve privacy in security-related data mining applications; we propose a QR Decomposition method for data distortion. We focus primarily on privacy preserving data clustering. As the distorted data occupies small amount of storage space, the memory requirement becomes low. Finally, we evaluate the effectiveness of the method in terms of misclassification error rate. Our experiments on several data sets reveal that the classification error rate varies as a result of security. However, the method has much less computational cost, especially when new data items are inserted dynamically.