Real Time Feature Convergence Measure for Efficient Discrimination for Transactional Data Set

Abstract

The problem of discrimination in transactional data set has been well studied. Numerous techniques has been recommended by various researchers but suffer to achieve higher performance. To handle this issue, a real time feature convergence measure based discrimination prevention algorithm is presented in this paper. The method first eliminates the noisy records by preprocessing the transactional data set. Second, the transactional data set has been grouped into number of clusters according to the pattern relevancy measure (PRM). Using the clusters generated, the the feature convergence measure (FCM) is computed for each item towards each cluster. The value of FCM is used to select a subset of items as sensitive one. Based on identified sensitive items, the method performs sanitization using probabilistic mapping scheme. The FCM algorithm supports the performance development of sanitization and discrimination prevention.