一个面向大规模数据库的数据挖掘系统

数据挖掘融合了数据库技术、人工智能和统计学,是目前的研究热点.为了能够集成当前数据挖掘的主要技术并使它们协同工作,在进行数据挖掘基本算法研究的基础上研制开发了一个数据挖掘系统--Golden-Eye.系统实现了在数据挖掘研究中的一些最新成果,集成了泛化、数据清洗这两个数据准备操作以及关联规则发现、例外规则发现、时序模式发现、分类器构造、聚类分析等基本数据挖掘操作,并实现了对挖掘操作的基本管理和结果的图形化显示.整个框架设计充分体现了系统的完整性、协调性和高效性:自底向上将存储控制模块、数据预处理模块、挖掘操作模块、挖掘库管理模块有机地结合在一起,在底层实现了对包括中间结果在内的数据的统一管理,在上层为用户提供了可视化的界面.实验结果表明,该系统能够在大规模数据库上成功地完成用户所指定的数据挖掘操作.

A Data Mining System for Very Large Databases

Data mining is a hotspot that combines the techniques in databases, artificial intelligence and statistics areas. On the basis of the research on some data mining algorithms and their implementation, a data mining system, Golden-Eye, is developed to incorporate primary data mining techniques and coordinate their operations. As the integration of several existing techniques including some improved algorithms as well as some newly proposed operations in data mining area, the system implements a wide spectrum of data mining functions such as generaliztion,data cleaning,association rele mining,exception rele mining,sequentil pattern mining, classification and clustering.By tightly integrating different functional modules such as storage management,data preprocessing,mining operations and mining base management,the system succeds in managing all kinds of data including midterm results uniformly and providing a user-friendly,visualized interface,which makes Golden-Eye a complete and efficient system with good perfmance.Experimental results show that the system can successfully fulfill the mining tasks specified by users on very large databases.