矩阵的奇异值分解在文本分类研究中的应用

Application of Matrix Singular Value Decomposition (SVD)to the Research of Text Categorization on Style

用向量空间的方法对文本进行分类是指通过对训练语料的学习，对每个类建立反映该类特征的向量，然后对每一个新的文本，提取其反映该文本特征的向量以确定其类别的过程。文章利用矩阵的奇异值分解理论，把被测文本的向量与所建立的标准向量(降维后)逐一比较,自动确定该文本的分类，经过开放式试验，获得了较好的效果。

By analyzing the training corpus, it can construct the vectors that reflects each existing class, and then create a vector that reflects the texts feature from each text for testing. With the theory of SVD, the authors compare the vector with each vector constructed to determine the style of text. It creates a good result with an open-experiment.