基于多层次概念提升的知识发现方法

A Knowledge Discovery Method based on Multi-level Concept Generalization

1 引言 数据库中的知识发现KDD(Knowledge Discovery in Database)，也称为数据挖掘(Data Mining，简称DM)，是近几年来随着数据库和人工智能发展起来的一门新兴的数据库技术。它从大量原始数据中挖掘出隐含的、有用的尚未发现的信息和知识(如规则、模型、规律、模式、约束等)，帮助决策者寻找数据间潜在的关联，发现被忽略的因素，因而被认为是解决现代社会“数据爆炸”和“数据丰富，信息贫乏”的一种有效方法[7,8]。

Concept generalization, which is also called attribute-oriented induction, is a KDD method widely used. Through concept generalization we can improve the abstract level of attribute. Thus we can get more succinct rule. But with increasing of the number of attribute and the more and more complicated concept levels,the traditional method based on the set theory becomes lower and lower efficient. We propose a new heuristic algorithm based on genetic algorithm in this paper,which seems to work well while dealing with large scale and complex problem.