An Evolutionary Approach using Transactional Impact Factor for Preserving Privacy of Quantitative Data

Extracting hidden useful knowledge from large collection of data is the definitive goal of data mining. But it may create serious threat to business and individual privacy. In this paper, a new method for preserving privacy of sensitive interval based quantitative association rule is proposed. Genetic Algorithm is employed to find the optimal intervals for quantitative rules without relying on support and confidence framework. Then, a mechanism is used to find the number of transactions to be perturbed based on the impact of sensitive rules and non sensitive rules each transaction supports. The proposed algorithm repeatedly modifies selected transactions thereby reducing number of modifications to the database. The main purpose of this method is to fully support the security of the database and to maintain the utility and certainty of mined rules at highest level. Experimental results show that the generation of is reduced by14% and Ghost Rules has increased by 17% than the previous work.