利用引用结构来提高科技文献分类的质量

Improving the Quality of Scientific Documents Classification by Using Citation Structure

科技文献的分类是科技文献数据库的重要组成部分,设计好的科技文献分类器是建立科技文献数字图书馆的重要任务之一.传统的文献分类法几乎都是基于文本的,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.本文利用科技文献的邻居文献所属的类这一信息结合科技文献的文本提出了一种协调的科技文献分类方法,并取得了较好的结果.

The classification of scientific document is a important component of scientific document database, a good scientific document classifer is a challenge of building scientific document digital library. Traditional classification methods of document are based-text, this make some documents which on the boundary of classes can not be classfied exactly. In fact, scientific document is one kind of semi-structure document, they have a lot of structural information ,which can be used to the classification of documents. This paper combine the class of one document's neighbor and its text gains a combined classification method of scientific document The experiment results mean that it is a good method.