摘要
关联规则挖掘作为一种大数据挖掘方法,被用于确定不同项目间存在的内在联系,并以大于某一阈值作为评判不同项间存在关联的依据。传统关联规则挖掘方法仅能建立不同项间的布尔型关联规则,存在硬化数据“尖锐边界”问题导致关联规则挖掘性能下降的缺点。为克服经典数据挖掘算法的这一缺陷,提出一种新型中智关联规则挖掘算法。基于专家知识对语言学术语进行量化预处理,得到不同指标的量化数据资料库;基于关联规则支持度定义计算不同指标项集合的支持度;通过考虑不同项间的隶属度、不确定度和非隶属度函数生成关联规则。将该中智挖掘算法与模糊挖掘算法进行对比,结果表明,该算法能够增加生成关联规则数量,有助于提高数据挖掘的准确性。
Association rule mining,as a big data mining method,is used to determine the intrinsic relationship between different items,and we use a threshold greater than a certain value as the basis for judging the correlation between different items.The traditional Boolean association rule mining method can only generate the binary rules among different items.However,the problem of hardening data “sharp boundary” leads to the performance degradation of association rules mining.To overcome this drawback,we proposed a neutrosophic association rule mining algorithm.Based on expert knowledge,the linguistic terms were preprocessed into the quantitative values to obtain quantitative databases of different indicators.The support degree of different index item sets was calculated based on the definition of association rule support degree.Furthermore,the association rules were generated by considering the membership,indeterminacy,and non-membership functions of different items.We compared the proposed algorithm with the fuzzy mining algorithm.The results show that our proposed can increase the number of generated association rules and help to improve the accuracy of data mining.
作者
梁凡
赵丽
Liang Fan;Zhao Li(College of Information Engineering,Nanning College for Vocational Technology,Nanning 530008,Guangxi,China;School of Software,Shanxi University,Taiyuan 030013,Shanxi,China)
出处
《计算机应用与软件》
北大核心
2019年第10期285-292,298,共9页
Computer Applications and Software
基金
2019年度广西高校中青年教师科研基础能力提升项目(2019KY1232)
关键词
大数据
数据挖掘
中智关联规则生成
模糊化理论
隶属度函数
Big data
Data mining
Neutrosophic association rule generation
Fuzzy theory
Membership function
