Research On Distributed Data Mining Systems Based On Web Service

For the following reasons, the original centralized data mining became more and more out of date:1. The data source need to be processed is distributed on the different computers in the networks.2. For the constrain of networks band, the privacy and safety of data, the incompatibility of systems, etc, it is not realistic to put all data source in a place (for example, the data warehouse) for centralized data mining. 3. More and more demands have addressed on the openness and easy accessibility. The distributed data mining technology was presented for the problems mentioned above. Presently, the two important matters in this field are that, design for suitable architecture of distributed data mining systems and corresponding distributed mining algorithms. This article introduced the latest technology for distributed component technology—Web services technology into distributed data mining field, and took some tentative efforts in solving the aforementioned two problems.In the beginning, the background for bringing distributed data mining, the status of research and research achievements, the existing problems, and algorithm for association rules were introduced. And then, the web services and related technology, and the advantages and disadvantages of web service technology were introduced, and the connecting point for web service technology and distributed data mining. And then a multi-platform, easy-extensible, suitable for distributed environment and web-based services distributed association rule mining algorithm FDM-GS (FDM with global site) were proposed. This algorithm adopted a new pruning strategy of candidate set and it can decrease the scale of candidate set and the networks information flow for collecting candidate set supporting counts. In addition, the detailed explanation for this algorithm was made with a practical example.