Michigan-based variable-length encoding of genetic algorithm for k-anonymization

K-anonymization is an effective method to protect personal privacy issues. Recently, genetic algorithm-based clustering approach has been successfully applied to the problem of k-anonymization. However, traditional genetic encoding has low efficiency and large information loss. This paper proposed a Michigan-based variable-length coding genetic algorithm, and the proposed approach adopts various heuristic strategies to select genes for crossover operation. Experimental results show that this method can further reduce the information loss and it is a new way to resolve the problem of k-anonymization.