A cryptographic approach for achieving privacy in data mining

Data governs the present world. Data mining has become a vital tool among many industries and organizations for effective decision making. Data mining involves extracting patterns from large data sets to form an understandable structure for future use. In recent years, sharing of data between organizations has taken place for their development. Even though sharing of data has an advantage, data which contains sensitive attributes pose a threat of exposing private information. Thus privacy of the sensitive attributes is of a greater concern. This work aims at preserving privacy of sensitive attributes by using two modified cryptographic techniques namely, Rail Fence and Vigenere Cipher algorithms. The unique feature of this approach is that the encryption matrix is formed from the original data itself. The modified data is generated based on this key matrix. From our experimental results it is evident that the original data cannot be inferred from the modified data.