IL-AdaBoost algorithm for XML document classification

An improved AdaBoost algorithm,IL-AdaBoost,for XML document classification is proposed.IL-AdaBoost uses frequent change of substructure with XML as the feature to build the decision stumps,then uses the decision stumps as weak classifiers,and improves AdaBoost algorithm.IL-AdaBoost is used to simulate new generation of XML documents through Poisson process,which reflects the characteristics of increase in XML documents with time,and updates the distribution of the sample to achieve incremental learning.IL-AdaBoost reduces the differences of basic classifier by sampling,and improves ensemble learning.