摘要
采用不确定规则形式表示发现知识和用户的领域知识,提出了发现规则相对于领域规则的前件一致度的概念及其计算方法,在此基础上给出了3种规则意外形式;分析了规则意外度的影响因素,给出了意外性的度量函数;研究了一种意外规则的挖掘方法,该方法不受产生规则的数据挖掘算法限制。实验结果表明,该方法能够有效地遴选出意外规则。
An important task for data mining is to discover interesting rules for the user. Unexpected rules shake user′s domain knowledge, so they are always interesting to the user. The discovered knowledge and user′s domain knowledge are expressed in uncertainty rule forms. The concept of rule’s antecedent similarity measure and its calculating method are proposed. Three types of unexpected forms are given. The factors which affect rule unexpectedness are discussed and functions to measure rule unexpectedness are given. A method for discovering unexpected rules is studied and it is not limited by data mining algorithm which produces rules. Experimental result shows that the method is effective.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2005年第3期381-385,共5页
Journal of Nanjing University of Aeronautics & Astronautics
基金
国家科技部基金(2002ED691036)资助项目。
关键词
数据挖掘
意外性
兴趣度
领域知识
data mining
unexpectedness
interestingness
domain knowledge
