Abstract:Since the automatic of Chinese word will bring the lack of information, we provide word segmentation according to lexical chunk as the unit. We divide such segmenting process into three sub-process: firstly, we segment text by means of Backward Maximum Matching. Second, we delete the stop-words from the segmentation result. At last, we count words mutual information and adjacency by the first time we segment words, and then, according to this counting result we can judge and sign the lexical chunk by relevant words. The experimentation shows that after the word combination, the lexical chunk bear much more feature information which shares a better effect of the process. It also has proved the effect of Feature Selection in Chinese Text Categorization and enhanced the capability of text classification.