A WSRF-enabled distributed data mining approach to association rules WEKA4WS -based

As a latest member in distributed computing technology family, the grid computing can play an increasingly important role with the progress of the DDM(Distributed Data Mining) technology in recent years. However, conventional data mining is not satisfied with the requirement due to the heterogeneous and distributed of the dataseis. Grid computing emerged as an important new field of distributed computing, which could support for distributed knowledge discovery applications. This paper has a try at combing the grid with web service in order to solve the problem of distributed association rules mining based on the research for the matrix theory, and achieves the distributed association rules algorithm by dint of Weka Library, presents a fast distributed association rules algorithm matrix-based, and also proves the correctness of the algorithm in theory. Finally it verifies the validity of the algorithm and the feasibility of the architecture with the distributed association rules based on WEKA4WS. This effective and fast algorithm shows sound extension, short time complexity, space complexity and small communication cost. To evaluate the efficiency of the proposed algorithms, a performance analysis of Weka4WS for executing distributed data mining tasks in different network scenarios are presented.