Secure Distributed Data Mining

Abstract

Data mining on a large dataset is a complex and time consuming task. The mining process on this large volume of data becomes slow, as it has to be done serially. The solution to the problem is to accelerate the mining process with the help of parallel or distributed approaches. Through mining, interesting relations and patterns between variables of large database can be observed securely using cryptographic techniques and the mining algorithms. This paper addresses the problem of secure distributed association rule mining over the horizontally distributed database. Security is the main problem in association rule mining projects. The solution to the problem is to accelerate the learning process with the help of parallel or distributed approaches. As mentioned earlier the performance of data mining algorithm can be enhanced from O(N) to O(N/k) with parallelism, where N = number of data instances and k=number of nodes [2]. There are several sites in the transaction. This system is based on distributed mining algorithm, K&C algorithm and AES algorithm. Distributed mining algorithm proposed here is the distributed version of apriori algorithm. The cryptographic technique is used to provide security in order to minimize the information shared in mining. With proposed method speed up is acquired while preserving the privacy of the data.