Research On Key Technologies Of Data Mining Grid

The main aim of this dissertation is to research the data mining architecture and the key technologies so that we can guide the design and construction of the Data Mining Grid (DMG) .Several challenges on the computation of the data mining are introduced in the field of telecom. This dissertation tentatively proposes the principle of data mining grid and addresses the emerging issues. The key technologies and problems of the DMG are analyzed. This dissertation proposes ideas and methods to solve those problems and present the prospect of DMG.In order to solve the chanllengs of the DMG, this dissertation studies several key technologies. Generally, the research works of this dissertation have six as following:●Survey of the theory and technologies for data mining and grid computing. After analysising the features of the classical algorithm and comparising the architectures of grid nowadays, the deficiencies of these grid architectures are presented. Indicates that the realization of the DMG is the right way to solve the problems of telecom data mining.●Proposed architecture of the data mining grid.By analyzing the requirements of the data mining grid and comparing the architectures of the grids, this dissertation proposed a DMG architecture. Then this dissertation descripts many features of the DMG. for the purpose of proving the correction and the practicability, and a paradigm is designed and the scene ananysis method is introduced in.●Proposed the parallel algorithms for various data minig algorithms, then general parallel method is proposed. The optimized schedule algorithm is realized. Then the parallel algorithm and the schedule algorithm are analyzed. This is a success of parallel algorithm for fulfilling for the data mining task, which can support the data analysis and the knowledge discovery for telecomm carrier businness intelligence.●Survey the meata-data criterions and the meta-data APIs, and analyze the importance and the reqirements of the meta-data for the DMG Then propose a meta-data model to suite to the need of the data mining grid system. For the purpose of the normalization and the interoperation between the computers, the XML Schema is used to descript the model. In order to show the practicalbility of the meta-data model, the instances are given.●The meta-data service architecture is proposed. Beside the design of the meta-data model, the mechanism of the management and the publication of the meta-data are presented as well. An instance using the web service shows the practicalbility of the architecture.●Designed and realized a DMG prototype system to prove the feasible and the practicalbility of the DMG. by running the workflows of some given business intelligence solution on the DMG prototype. The feasibility and the efficiency are proved.