Multiclass Text Classification A Decision Tree based SVM Approach

This paper discusses about combining Support Vector Machine and decision trees for multi class text classification. Support Vector Machines are trained on each class at each level of the tree and the SVM which is more successful in predicting a class at that level is selected as the decision in that node. Thus a tree is constructed with different SVM in each node. And the tree constructed is used for classifying the multiclass text. Results had shown that this method works comparably better than the other classifiers like simple SVM and Naïve Bayes.