The Design And Implementation Of The Component Library In Zmining Data Mining System

With the development of computer and information technology, all walks of lifefilled with large amounts of data, how to analyze and explore the relationship and ruleof data, to detect the potential impact of data，has become a research hotspot.Data mining is the theory and methods which integrated computer science,statistics, artificial intelligence and related disciplines, it combined with knowledge ofthe relevant industry background and historical data to analysis and explore theunknown but potentially useful knowledge and patterns. As the application field ofdata mining theory, data mining system has played a very important role in thedevelopment of data mining technology.Zmining data mining system support mainstream data preprocessing and datamining operation, there is no dependency between data mining processing unit anddata containers, it is convenient for user to preprocess and data mining of theheterogeneous data sources. The data mining component library is an important partof data mining system. The design of component library based on the thought of theEclipse plug-in architecture, achieve the purpose of the seamless connection betweeneach module. The component library has a good scalability and maintainability,suitable to build "on-demand and become" data mining tasks. In the system, the basicprocessing units of data mining are called nodes. Workflow fragments through theextension point, which is provided by the component library, are packaging tocorresponding components. When some workflow fragments are frequently andalways required in the specific process of data mining, users can directly use thecomponents. These components through a large number of experiments andverification, can directly apply to the relevant data mining tasks. On the one hand iteliminates the node configuration, and on the other hand it also improves theefficiency of data mining in a certain extent.