Research on Improvement of Bagging Chinese Text Categorization Classifier

In text categorization ensemble learning is one of the methods for improvign the predictive power of classifier.Bagging algorithm is a popular ensemble learning now. Aiming at the problem that weaker classifiers of Bagging have the same weights,an improved Bagging algorithm is developed. The confidence of weaker text classifiers are gained through the result of weaker classifier and the weights of voting is obtained by confidence. The algorithm is applied in Attribute Bagging algorithm to design a Chinese text classifier. Using kNN as the weaker classifier model,which classify news corpus of Sogou lab. The result of experiment shows that this algorithm performs better than Attribute Bagging with more accuracy.