关键词: Bigram文本特征, &chi, ²统计量, 互信息, 维吾尔语

Abstract: Text representation is the most important phase in automatic text categorization. In the vector space model based text representation, the selection of feature granularity has the direct impact on the text categorization performance. The word features don’t have the good representative power to represent the Uyghur texts in text categorization. To solve this problem, the CHIMI based Uyghur Bigram extraction method is proposed and the Uyghur text categorization experiments are conducted using support vector machine algorithm based on the extracted Bigrams as text features. The experimental results show that the Bigram based Uyghur text categorization achieves higher classification precision and recall compared to the word based categorization and experiments demonstrate the effectiveness of the proposed algorithm.

Key words: Bigram text feature, χ² statistics, mutual information, Uyghur Language