基于XML结构相似性的名老中医病例聚类分析  

Structure-similarity cluster analysis on herbal instances based on XML

作　　者：班晓娟[1] 宁淑荣[1] 马骥[1] 尹怡欣[1] 张德政[1] 

机构地区：[1]北京科技大学信息工程学院,北京100083

出　　处：《计算机应用研究》2008年第2期365-367,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60503024;60374032);国家"十五"科技攻关计划资助项目(2004BA721A01H07)

摘　　要：针对名老中医病例的结构特点,设计了一种基于模拟退火的聚类算法对数据库中病例聚类进行全局优化。病例聚类时,根据一般意义的树间编辑距离,提出一种用于判断XML描述的病例之间是否相似的度量(称为XML编辑距离)。利用XML编辑距离,可将XML数据间相似性度量的时间复杂度限制在多项式级,且能保持病例的XML描述文档的节点语义信息和节点间的祖孙嵌套关系。最后,在Tamino数据库上进行实验,结果证实了基于模拟退火的病例聚类算法在名老中医数据挖掘实践中的可行性和有效性。According to characters of herbalist instances, the article designed a clustering arithmetic which based on simulating anneal to optimize the herbalist instances in database. It provided a measurement which consulted usual editing distance between trees to estimate similar degree between instances described by XML（ named for XML editing distance）. If made full use of XML editing distance, the time complexity which calculated similar degree between XML data could keep in multinomial level, furthermore, the semantic of nodes in document described by XML and the nested relationship among nodes could be preserved, Finally, the test performing in Tamino database gets a good result and proves that it is a feasible and effective clustering arithmetic.

关 键 词：名老中医病例 可扩展标记语言描述文档 可扩展标记语言编辑距离 聚类算法 模拟退火算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]