一种协调的科技文献分类方法

科技文献之间的相互引证关系反映了一种科学交流活动,显示了科学文献之间(甚至是学科之间)的内在联系,而通过追溯文献之间的这种关系,可以改善和提高传统的基于内容的科技文献的分类的方法。论文利用有相互引证关系,有同引关系,以及有耦合关系的两篇文献一般是属于同一类的这一特点,提出了文献之间的引用相似度,同引相似度,耦合相似度这三个概念,再利用这三个概念生成了文献之间的“结构相似度”,并将它用于K-NN分类法中得出一种基于结构的分类法。最后,论文将这种基于结构的分类法和基于内容的NaveBayes分类法结合起来提出了一种新的协调分类法。

A Combing Classifying Method of Scientific Documents

Mutual citing between scientific documents reflects a kind of scientific exchange activity,shows the internal relations between scientific documents,even the internal relations between different fields.By tracking this relation be-tween documents,we may improve traditional scientific documents classifying method based on content.Making best use of the feature that two documents having mutual citing or co-citation or coupling relations belong to the same category,we propose three concepts between documents,i.e.,citing similarity,co-citation similarity,coupling similarity.Then we cre-ate structural similarity between documents by the aid of these concepts.The structural similarity is put into K-NN classifying method and a classifying method based on structure is formed.Finally,we propose a new reconciled method by combining classifying method based on structure with classifying method based on Naive Bayes.