Based on Vector Space Model(VSM) and Nave-Bayes(NB), completed a multilayer and multi-classification text categorization system. Introduce detailedly four modules: words’ segmentation and frequency statistics, calculating between classifications’ and document, emendating the veracity of parent-class by emendation of subclass, judging whether document has multi-classification and multi-label. Text representation based on Vector Space Model has 89.7% MicroF1 of parent- category, 77.8% of sub- category; text representation based on Nave-Bayes has 67.6% MicroF1 of parent- category, 66.5% of sub- category.