Chinese Question Classification Based on Ensemble Learning

In this paper, a method of Chinese question classification based on ensemble learning is proposed. Words and bi-gram are extracted from question as classic features and feature classifiers are constructed. The classifiers are kinds of simple classifier and their generalization ability are not strong enough, but they are a qualified base learner for ensemble because of their low computational cost, and the generalization feasibility improved with the help of ensemble learning. We translate and modify the UIUC question set and TREC2001 question set as Chinese question set. Experimental results on the corpus show that the proposed method can achieve good performance, the classification precision reaches 87.6%, under the fine grained question types.