本文阐述了自动分类中的仿人思想 ,并用分类域模型来描述中文信息自动分类中的分类法 ,通过计算类别特征项在分类域中的Hamming距离 ,对类别特征项依据其在分类域中的类别分布进行聚类 ,从而实现对向量分类法中的特征向量维数的压缩 ,并进一步构造用于中文自动分类的知识库。 This paper describes the human-like thinking of automatic classification, and uses the classification domain model to describe the classification of Chinese information automatic classification. By calculating the Hamming distance of the category features in the classification domain, Class distribution, so as to realize the compression of the dimension of the eigenvector in vector classification and to further construct a knowledge base for Chinese automatic classification.