Abstract: The k-anonymity problem has recently drawn considerable interest from research community and a number of algorithms have been proposed. We are also considered the problems of existing methods and addressed such limitations with the aid of our proposed k-anonymization technique. We have intended to propose using a clustering based K-Anonymous method. The primary goal underlying our approach is that the k-anonymization problem can be considered as a clustering problem. Intuitively, the k-anonymity requirement will be generally transformed into a clustering problem, where it is required to discover a set of clusters each of which contains at least k records. Moreover, our proposed research will be reduced the Information Loss. The records will be initially collected and which will be analyzed for two different attributes: Numerical attributes; categorical attributes. For these two kinds of attributes, we will then find the distance of records, separately. Followed by the distance calculation, each record will be evaluated and the Information Loss (IL) in each and every record will be obtained by our proposed research. This will be performed based on the anonymity value for the input dataset. Now, we will obtain the loss of information in each record and from this we can cluster out the minimum IL record. The clustering will be done by an Adaptive Particle Swarm Optimization based Fuzzy C-Means (APSO-FCM) Clustering algorithm. Fuzzy C-Means (FCM) is one of the clustering algorithms used to make a group of data into clusters, in which one data can be allocated for two or more clusters. In FCM, the objective function will be optimized with Particle Swarm Optimization (PSO) algorithm. The PSO algorithm imitates the social characters shown by swarms of animals. In this algorithm, a point in the search space which is a possible solution, is called a particle. The group of particles in a specific iteration is called swarm. While looking out for food, the birds are either scattered or go collectively before they find out the place where they are able to locate the food. While the birds are on the search for food moving from one location to another, there is often a bird which is able to smell the food effectively, in other words, the bird is discernible of the location where the food is likely to be found having superior food resource data. As they tend to convey the data, particularly the excellent data at any time while looking for the food from one location to another, attracted by the excellent data, at the end, the birds will throng at the location where there is strong possibility for locating food. Thus, the clustering of nearly minimum of IL records will be gained by our proposed research. The proposed approach will be implemented in MATLAB and planned to be evaluated using various databases. Our proposed research will be made better performance evaluation results with reduced Information Loss.
G. Chitra Ganabathi and P.Uma Maheswari, 2016. Privacy Preserving K-Anonymization Clustering Approach for Reducing Information Loss. Asian Journal of Information Technology, 15: 1531-1538.