基于摘要文本的专利快速自动分类方法

The Rapid Automatic Categorization of Patent Based on Abstract Text

随着我国专利申请数量的迅猛发展,对受理专利实现快速分类的需求也越来越迫切。作为专业性知识极强的科技文献,专利自动分类的正确率远高于普通文本的分类效果。基于专利摘要进行快速自动分类的方法,充分考虑专利类别的层次结构特性,建立不同层次的类别特征中心向量,采用经典的类中心向量分类算法实现专利的快速自动分类。实验结果表明:该方法在专利不同层次上的分类效果明显,时效性大大提高,能够满足专利快速分类的实际需求。

With the rapid development of the number of patent applications in China, the demand for the rapid categorizationof the accepted patent is becoming more and more urgent. In terms of professional document of science and technology, the correctrate of patent automatic categorization is much higher than the ordinary document. The rapid automatic categorization of patent basedon text gives full considerations to the hierarchical structures of the patent classes, establishes the center vector of class fea-tures at different levels, and uses the classical algorithm （the center vector categorization method） to realize the rapid categoriza-tion. Experimental results show that the method works well at different levels, and the effectiveness is greatly improved, which canmeet the practical requirements of the rapid categorization of patent.