Hierarchical Text Classification Incremental Learning

To classify large-scale text corpora, an incremental learning method for hierarchical text classification is proposed. Based on the deep analysis of virtual classification tree based hierarchical text classification, combining the two application models of single document adjustment after classification and new sample set learning, a dynamic online learning algorithm and a sample set incremental learning algorithm are put forward. By amending classifiers and updating the feature space, the algorithms improve the current classification models. Hierarchical text classification experiments on Newsgroup datasets show that the algorithms can enhance the classification accuracy effectively and reduce the storage space and the learning time cost of the history sample datasets.