Study on Short Text Classification with Integrated Algorithm

With the rapid growth of the number of short text, how to effectively realize the automatic classification of short text is needed to be solved in the information domain. According to the characteristics of short text, this paper proposes Bagging_NB & Bagging_BSJ, which are two classification algorithms based on the improvement of current integrated classifiers. Traditional classifier NB, SVM, J48 are used as the basis classifiers to train the classification models. Compared with several individual classifiers, our methods have excellent results in a variety of classification evaluation indexes.