一种协调迭代的科技文献分类方法  

A Combined Iterative Classification Method for Scientific Documents

作　　者：檀林[1] 张永奎[1] 

机构地区：[1]山西大学计算机与信息技术学院,山西太原030006

出　　处：《山西大学学报（自然科学版）》2004年第3期249-253,共5页Journal of Shanxi University(Natural Science Edition)

基　　金：山西省回国留学人员基金(2002004)

摘　　要：科技文献的分类是科技文献数据库的重要组成部分.传统的文献分类法几乎都是基于文献内容的,这样会使一部分处于类边缘的文献不能准确地分类,事实上科技文献是一种半结构化的文献,它们包含的很多结构信息可以用到文献的分类中.利用科技文献的毗邻文所属的类这一信息结合科技文献的内容提出了一种协调迭代的科技文献分类方法,并给出了该迭代算法的收敛性证明.最后对算法及实验结果进行了分析,表明该方法比传统的分类方法性能有所提高.The classification of scientific document is an important component of scientific document database.Traditional classification methods of the documents are based-content,however,that make some documents which on the boundary of classes cannot be classified exactly.In fact,scientific document is one kind of semi-structure document,which has a lot of structural information that can be used to the classification of documents.Combining with the class of document's neighbors and its text can gain a combined iterative classification method of scientific documents.The justifying of the iterative method was also given.Finally,the arithmetic and the experiment were analyzed.The results showed it was better than the traditional classification methods in performance.

关 键 词：协调迭代 科技文献 分类方法 毗邻文献 引文 

分 类 号：G254.1[文化科学—图书馆学] TP18[自动化与计算机技术—控制理论与控制工程]