Secure Mining of Generalized Association Rules from Horizontally Distributed Databases

We consider the problem of secure mining of generalized association rules from horizontally distributed databases. Given a large horizontally distributed database of transactions, where each transaction consists of a set of items and taxonomy on the items, we find associations between items at any level of the taxonomy. Generalized association rule mining technique has discussed in many papers. But, in this paper we discuss about secure mining of generalized association rules from horizontally distributed databases or homogeneous databases. For that purpose, we use the same privacy preserving distributed mining concepts discussed in paper ^[1] with the generalized association rule mining technique called ‘cumulate’ algorithm discussed in paper ^[2]. The main privacy preserving parts of the protocol in paper ^[1] are two secure multi-party algorithms called UNIFI and SETINC. Our proposed protocol is based on Fast Distributed Mining (FDM) algorithm. FDM algorithm is an unsecured distributed version of Apriori algorithm. It offers enhanced privacy, simplicity and efficiency.

Keywords: Privacy Preserving Data Mining, Horizontally Distributed Databases, Generalized Association ruldes, Frequent Itemsets..