关键词: 文本分类 特征值提取 自然语言处理

Abstract:

Based on the characteristics of Chinese texts, we propose a method of Chinese text eigenvalue extraction using multiple heuristic rules in this paper.  The eigenvalues of texts are extracted according to many aspects. In the extraction we substitute the traditional word with a kind of new synonym conce ption as the units of eigenvalue and consider not only the appearance rates of words but also the semantic information in the text. And, the model and a  lgorithm of eigenvalue extraction are provided in this paper. Finally, we present the results of experiments comparing with traditional extraction metho ds using the appearance rate of words in the text, which illustrates that the method of this paper improves the correctness rate of text categorization  and reduces the dimensions effectively.

Key words:  （text categorization, eigenvalue extraction, natural language processing）