Study of Tibetan text categorization based on ensemble learning classifier

Based on the text feature and syntatic structure of Tibetan, this theory mainly focuses on categorization of Tibetan through ensemble learning classified method. Combine KNN and Naive Bayesian to build a basic classifier on account of three character subsets over category of word character. And then take weighted calculation of basic classifier. For the last step, calculate the weight of basic classifier via gradient descend and reach final result of categorization. During experiment, choose recall ratio, precision ratio as well as other evaluation function to analyze and evaluate KNN, Naive Bayesian and text categorization. Conclusion from experiment shows that precision of calculation of Tibetan text based on ensemble learning method has been greatly improved.