Research of the mining algorithm based on distributed database

Data mining acquires knowledge and rules that is connotative, unknown and having potential value for decision-making from large databases or data warehouses. Existing association rule mining algorithms and modules cater to a centralized environment, such as database or data warehouse. With the development of distributed database and network technology, they don´t meet the needs of mining rules from distributed data sets. In this paper, we provide a mining model based on the distributed database and a corresponding effective mining algorithm. Using association rules of market basket analysis, integrate every database file, then get the mining result, and make a further mining upon the mining method, transport the rules which are not fit with the requirements back to each distributed station to make a more accurate mining process, thus avoiding the frequent network communication. This algorithm can reduce frequent communication burden, they has an distinguish virtue in parallel arithmetic computing and asynchronous operation & heterogeneous mining.