摘要
为了为姜黄属植物分类与中药鉴定提供新的参考,以姜黄属(Curcuma L.)的31个材料为研究对象,把47项分类性状(指标)分解为28个定性指标和19个定量指标,将28个定性指标进一步分解为二元性状指标和三元性状指标,然后各自进行粗糙集理论的处理和聚类,把聚类结果按交集分类的办法形成定性指标综合分类的结果;将19个定量指标进行粗糙集理论的处理和聚类,把聚类结果与定性指标的分类结果按交集分类的办法形成20项指标综合分类的结果。结果表明,将定性性状指标分类集与定量性状指标分类集进行交集分类,姜黄属植物的31个材料可分为10个类群(种);这与前人采用47项指标一次性完全聚类的结果存在差异,与传统分类也不完全相同。说明粗糙集理论对于分类指标的处理,一方面去掉了对分类没有显著作用的指标,减少了运算;另一方面减少了不必要指标对分类的影响,使分类更为准确。
To provide new evidences for the classification and medicinal identification of Curcuma L.31（OTU） kinds of Curcuma L.were studied in this paper.47 taxonomy characters were decomposed to 28 qualitative index and 19 quantitative index,and the qualitative characters were further decomposed to two-state characters and three-state characters,then treated by rough-set theory and clustered,respectively.The results of cluster analysis were divided into qualitative index synthesis taxonomy by intersection taxonomy.19 quantitative characters were treated by rough-set theory and cluster analysis,and the results were classified into 20 kinds of character synthesis taxonomy by intersection taxonomy,then synthesized into 10 major kinds.The results were compared with principal 47 taxonomy characters by component and cluster analysis for made by Xiao xiao-he.The study showed that it was reasonable and accurate to decompose taxonomy character,and combining rough-set theory with cluster analysis was significant.
出处
《西北农林科技大学学报（自然科学版）》
CSCD
北大核心
2007年第11期203-212,共10页
Journal of Northwest A&F University(Natural Science Edition)
基金
南阳市科技攻关项目(2005PT064)
南阳师范学院资助项目
关键词
粗糙集
聚类分析
姜黄属
数值分类
rough-set theory
cluster analysis
Curcuma L.
numerical taxonomy
