A Distributed Rule Mining Algorithm Based on Fp-Tree
A Distributed Rule Mining Algorithm Based on Fp-Tree

Summary/Abstract: Mining global association rules from a distributed system is a challenging task. A solution must be not only computationally efficient but also scalable to potentially very large distributed systems. There have been some distributed association rule mining algorithms reported in the literature, all of which are based on Apriori. Using these algorithms, the communication overhead increases as the number of local databases increases. In this paper we propose a scalable algorithm, called DFPT, which is based on FP-tree. One important characteristic of DFPT is that communication cost scales primarily with the size of local databases and not with the number of local databases. So, increasing the number of local databases in the whole system does not significantly increase the communication cost. DFPT is especially suitable for a system that consists of a very large number of local databases, each of which is of moderate size, such as a network of a large number of small computers or a peer-to-peer system.

• Details
• Contents