摘要
给出了一组关于分类问题的自足而严密的形式化描述,并运用精确和覆盖两种准则,将类的特征明确划分为充分特征和必要特征。在此基础上,结合“约简”方法和“聚焦”机制,提出了一种新的分类规则提取算法。使用该算法,能从分类信息系统中提取出明确的分类规则。这些规则不仅包含了分类信息系统中类的某些潜在的充分特征和必要特征,而且比传统的归纳法所提取的规则更能反映专家的“聚焦”思维方式。最后给出了运用该算法的示例。
First,a group of self-completed and rigorous formal descriptions about classification problem is presented.By means of two kinds of criterion,i.e.,accuracy and coverage,the characterizations of class have been definitely divided into two parts,the sufficient ones and the necessary ones.On such basis,combining reduct method with focusing mechanism,a new algorithm for classification rules extraction is proposed.Using this algorithm,one may from classification information system extract crisp rules,which not only consist of some potential sufficient and necessary characterizations of classes in the classification information system,but also reflect more obviously and accurately experts＇ thinking way of focusing mechanism than the traditional induction approaches.At last,an example is presented to illustrate the proposed algorithm.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第9期150-153,165,共5页
Computer Engineering and Applications
基金
国家自然科学基金资助项目(编号:60373000)
关键词
知识获取
规则提取
粗糙集
机器学习
数据挖掘
knowledge acquisition, rules extraction, rough set, machine learning, data mining
