Parallelization of a novel frequent itemset hiding algorithm on a CPU-GPU platform