摘要
现阶段关于方剂配伍规律的研究主要用到的是数据挖掘技术,包括分别以分类和聚类为主的研究模式和以关联规则挖掘为主的研究模式,这两种研究模式作为研究中药方剂的重要技术取得了一系列可喜的成果,但是它们在揭示中医理论体系复杂性和内隐性方面还有不足之处.提出一种新的复杂网络的模型来探索中药方剂配伍规律,并用改进的COPRA算法对构建的复杂中药方剂网络进行社团发现,最后通过用自定义的模块密度来衡量所发现的社团紧密程度,以及对发现的社团与关联规则挖掘算法挖掘出的最大频繁项集进行比较,发现该模型及算法有很好的效果,具有一定的实用性.
Recently,data mining techniques are mainly used in the research of opening-out-compatibility relationships of prescriptions,including techniques-adopted models based on classification or clustering and models based on association rule mining.As most important techniques,these two research models have made lots of promising results on traditional Chinese medicine formula（TCMF）,but they have disadvantages in revealing the complexity and implicit of the TCMF.We present a novel model based on complex network to explore the compatibility relationships of prescriptions.Then we detect the communities in TCMF network,using the improved algorithm of COPRA which overcomes the defect of label shocks in the classic COPRA algorithm.At last we take module density function,which is defined by ourselves,to measure closeness of communities and we compare communities with largest frequent itemsets,which come from association rule mining algorithm.Our model and improved COPRA algorithm show much better performance and have a certain practicability.
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第4期483-490,共8页
Journal of Nanjing University（Natural Science）
基金
国家973项目(2011CB505300)
关键词
中药方剂网络
COPRA算法
模块密度
traditional Chinese medicine formula（TCMF）network
COPRA algorithm
module density
作者简介
通讯联系人，Email：chjwang@niu．edu．cn
