基于支持向量机方法的中文组织机构名的识别

Chinese organization names recognition based on SVM

在应用基本的支持向量机算法的基础上,提出了一种分步递增式学习的方法,利用主动学习的策略对训练样本进行选择,逐步增大提交给学习器训练样本的规模,以提高学习器的识别精度。实验表明,采用主动学习策略的支持向量机算法是有效的,在实验中,中文机构名识别的正确率和召回率分别达到了81.7%和86.8%。

This paper introduced SVM-based method, which used active leaning strategy to incrementally select new instances to be labeled and included in its training set to form an incremental course of learning. The result of the test show that the method is efficient and the precision and recall of Chinese names recognition achieved are 81.7% and 86. 8% respectively.