【作者】 钟响；

【导师】 张洪伟；

【作者基本信息】 四川大学 ， 计算机软件与理论， 2006， 硕士

【摘要】 知识发现是从数据库中抽取和精化新的模式。信息系统是数据库的抽象描述,是一个具有对象与属性关系的数据库。信息系统的知识发现问题本质上是按照属性特征将对象进行分类的问题。面对巨型数据库系统,以及数据库系统表现的多样性,需要能够识别出正确、新颖和有潜在应用价值的模式。基于这种思想,本文的工作首先是引入了集值信息系统以及在信息系统上进行知识发现的工具——包含度。知识获取是从大量属性中寻求某些规则,而规则的前件与后件其实是一种包含关系,两个规则是协调的,是指两个规则前件的相似度不超过两个规则后件的相似度,因此,可以用包含度建立两个规则之间的协调度,作为属性重要性的一个度量指标。有了集值信息系统和包含度,就已经可以开始进行知识发现的工作了,但是这其中还有问题存在。本文研究了集值信息系统的知识发现过程,发现由于集值信息系统本身的不足使得包含度的计算过于粗略,这造成了知识发现精度不高,显得其不适合解决不确定性程度较高的问题,从而影响了其知识发现的应用价值。为解决这个问题,本文又引入了灰色系统理论,这一研究数据不确定性的利器。灰色系统理论是研究少数据不确定性的理论,即灰色系统理论是针对既无经验,数据又少的不确定性问题,也就是“少数据不确定性”问题提出的。在研究中,结合集值信息系统的特点,本文选择了灰关联度,一种描述序列之间联系紧密度的数量表征,它具有整体性测量的特性,来改善集值信息系统知识发现中包含度的计算。在完成理论部分的工作之后,本文在实践领域也展开了工作,将本文的研究成果应用于SCM更多还原

【Abstract】 Knowledge discovery is a process to abstract and refine a new mode from database. Information system is an abstract description of database which is also a database having the relationship of object and attributes. The knowledge discovery of information system essentially is the problem that how to classify the objects according to the characteristic of attributes. We want to identify the mode of correctness, novelty and latency valuable application though we are up against the problems of gigantic database and the multiformity that they presented. Based on this idea, in this paper the sets information system and inclusion degree, the tool of knowledge discovering in the information system had been introduced in firstly. The gain of knowledge is to seek some regulations from mass attributes and actually the relation between the antecedent and consequent of regulation is inclusion. And the similar degree of antecedent of the two regulations is not beyond which of consequent of the two regulations then it is said that the two regulations are concerted. So we can set up the correspond degree between the two regulations as a measurement index for measuring the importance of attributes. With the help of sets information system and inclusion degree, we can start working on knowledge discovery; however, there is still the existence of problems. In this paper the process of knowledge discovery of sets information system had been researched. And it is found that because of the shortness of sets information itself the computation of inclusion degree is rough. Then it leads the result that the precision of knowledge更多还原