Improved Genetic Algorithm Approach for Sensitive Association Rules Hiding

Association rule mining is interesting area of data mining research which discovers correlations between different item sets in a transaction database. Efforts have been made for efficient hiding of sensitive association rules, but these techniques do not consider the consequences such as loss of information, lost rules and increase in ghost rules production. In this paper, we propose improved genetic algorithm architecture with a new fitness function for hiding sensitive rules by reducing loss of information, lost rules and generation of ghost rules. Different datasets have been used for experimental analysis. The results show the superiority of our work over the existing techniques.