A Fast and Stable Clustering Algorithm for K-Anonymity Model
Tung-Shou Chen, Jeanne Chen, Chung-Kuang Chen, Yuan-Hung Kao · ASME Press eBooks · 2011
During the same time as data mining, it may lead to leakage of personal privacy issues; k-anonymization algorithm is usually used to protect personal privacy, but they are very complicated algorithms, when k-anonymity is a NP-hard problem, k is equal or greater than 3. Traditional clustering algorithm is used to improve the computational efficiency of k-anonymization, but it takes a higher time cost. In order to improve the abovementioned problems, we introduce a fast and stable clustering algorithm, to improve the k-anonymous operation and receive better results. The method of time complexity is only O(n*q), better than the One-Pass K-means Algorithm of O(n2/k), where n is the sampling number of data, q is the number of quasi-identifier and k is the number of anonymities. Proven via the experiment, and with k=50 as the example, this method execution time only takes 16 seconds, the One-pass K-means Algorithm execution time is 120 seconds. This study provides a fast and stable result of the k-anonymity algorithm.