摘要
围绕数据对象与用户需求之间的关联特性,该文将"本体论"基本理念引入聚类分析过程,定义"本体核"和"客体数据"来表述数据对象在数据空间中的动态分布,并通过弱(强)量本体核构建数据对象间的相互关系.同时,利用"本体核"中客体数据的频数及其同频异值数构造聚类分析直方图,分别使用两种控制聚类合理性函数来确定直方图门限值的预处理算法,获取数据对象的最终聚类归属,并通过应用示例验证方法的可行性和有效性.
According to the connective characteristic of data and demands within data space, the principle of ontology is introduced into the process of cluster analysis. The pinpoint existence modality and dynamic distribution of data are described on the definition of ontic kernel and object data. Interrelation among the data can be studied with weak ontic kernel and strong ontic kernel. The cluster analysis histogram is constructed on the time of object data and different value on the same time. For different histogram statement, the algorithm that can process the limited gate value with rational function of controlling cluster is presented to determine the ultimate generic cluster. An application illustration is used to demonstrate feasibility and validity of the cluster preprocessing on ontic kernel and histogram （CPOKH）.
出处
《上海大学学报（自然科学版）》
CAS
CSCD
北大核心
2008年第1期19-25,共7页
Journal of Shanghai University:Natural Science Edition
基金
上海市高校科学技术发展基金资助项目(04AB29)
关键词
本体核
直方图
合理性函数
聚类预处理
ontic kernel
histogram
rational function
cluster preprocessing
