A Technique to Mine Clusters Using Privacy Preserving Data Mining
J. Anitha, Raghuraman Rangarajan · International Review on Computers and Software (IRECOS) · 2013
In recent years privacy preserving data mining problem has gained considerable importance, due to the vast amount of personal data about individuals that are stored at different commercial vendors and organizations. Privacy preserving clustering is not intensively studied as other data mining techniques, such as rule mining, sequence mining, etc. In this paper, we obtain privacy by anonymization, where the data is encrypted from the original data, along with the secure key and the secure key is obtained by the Diffie Hellman key exchange algorithm. In order to perform clustering on the anonymize data Fuzzy C Means clustering algorithm is used. The Fuzzy C means clustering algorithm is suitable for clustering data where the boundaries are ambiguous. However in this paper initially distance matrix is calculated and using which similarity matrix and dissimilarity matrix is formed. Similarity matrix calculates the similarity among the data point with the cluster centroids and dissimilarity matrix calculates the dissimilarity among the data point with the cluster centroids. The membership matrix is constructed from the above matrices is used to cluster the anonymized data. The experimental result of the proposed algorithm is compared with K-Means algorithm in terms of running time, memory usage, and accuracy and it is proved that the efficiency of the proposed algorithm is better in terms of accuracy and execution time