【摘要】 从自然语言的角度考虑词性选择,同时从统计学角度考虑删除文档频率过低的特征词,从而避免产生维数灾难,通过考查类别本身特征和类别之间的关系来提取类别特征向量,采用传统夹角余弦公式考查文本与类别的相似度,实现一种过程简单,易于理解且分类效果不错的文本分类系统。更多还原

【Abstract】 From the view of natural language,POS selection is carried out;and from the view of statistics,the low frequency features are deleted.The relationship between the categories and their own features are examined to extract category features vector.Angle Cosine formula was used to examine the similarity between document and the category.A document classification system which is relatively simple,easily understood and works well has been presented.更多还原