An Effective Hybrid Machine Learning Approach for Coreference Resolution

We present a hybrid machine learning approach for coreference resolution. In our method, we use CRFs as basic training model, use active learning method to generate combined features so as to make existed features used more effectively; at last, we proposed a novel clustering algorithm which used both the linguistics knowledge and the statistical knowledge. We built a coreference resolution system based on the proposed method and evaluate its performance from three aspects: the contributions of active learning; the effects of different clustering algorithms; and the resolution performance of different kinds of NPs. Experimental results show that additional performance gain can be obtained by using active learning method; clustering algorithm has a great effect on coreference resolution’s performance and our clustering algorithm is very effective; and the key of coreference resolution is to improve the performance of the normal noun’s resolution, especially the pronoun’s resolution.