New Similarity Rules for Mining Data