In our era Knowledge is not "just" information anymore, it is an asset. Data mining can be used to extract important knowledge from large databases. These days, it is often the case that such databases are distributed among several organizations who would like to cooperate in order to extract global knowledge, but at the same time, privacy concerns may prevent the parties from directly sharing the data among them. The two current main methods to perform data mining tasks without compromising privacy are: the perturbation method and the secure computation method. Many papers and published algorithms are based on those two methods. Yet, both have some disadvantages, like reduced accuracy for the first and increased overhead for the second.

In this article we offer a new paradigm to perform privacy-preserving distributed data mining without using those methods, we present three algorithms for association rule mining which use this paradigm, and discuss their privacy and performance characteristics.