An approach to mining association rules in horizontally distributed databases with anonymous ID assignment

Data Mining is the technique of automated extraction of interesting data patterns used to represent knowledge, from the large data sets but sometimes these datasets are divided among various parties. Association rule mining is a popular mining technique that identifies interesting correlations between database attributes. In this paper, proposed a protocol Privacy Preserving Fast Distributed Mining (PPFDM) for association rules mining in horizontally distributed databases which is based on the Fast Distributed Mining (FDM) algorithm. FDM is an unsecured distributed version of the Apriori algorithm devoted to generate a small number of candidate sets and considerably cut down the number of messages to be passed at mining association rules. PPFDM adopts two major ideas: one that computes the union of private subsets that each of the interacting player holds and another that evaluate the inclusion of an element held by one player in a subset held by another. An implementation of a PPDM algorithm is developed in Java framework and performance results are presented for synthetic data generation and association rules as well as indexing is provided to the user. It is simpler and significantly more efficient in the matter of communication rounds, communication cost and computational cost.