An integrated method for mining association and sequential rules in distributed databases

Association and sequential rules have received much attention in both academia and in practice for quite some time. A number of algorithms have been proposed to improve the efficiency of mining decision rules. However, these centralized algorithms have been criticized for their inapplicability to distributed environments. Although many sophisticated distributed algorithms for mining association and sequential rules have been proposed in recent years, they are not supported by most popular software programs; this decreases the possible applications of decision rules. In addition, past decision rules have only focused on the rules for which items exist. Some important decision rules might be lost if suitable items do not exist. In this paper, we propose an integrated framework to combine centralized algorithms, such as the Frequent Pattern Growth (FP-Growth), Sequential Pattern Discovery using Equivalence classes (SPADE), and rough set algorithms, to mine decision rules in a distributed environment. In addition, our method finds some significant rules that other algorithms cannot. The experiments also demonstrate that the proposed method is well suited to finding association and sequential rules in a distributed environment.