基于语义角色的中文时间表达式识别

摘要

This paper proposed a new Chinese time expression recognition method with a variety of common features plus se-mantic roles features according to the characteristics of Chinese time expression and using the feature vector to train a CRFs-classifier. When recognizing the Chinese time expressions, this method achieved the Fl measure of 85. 6% which was in-creased by 5. 2% compared with no semantic role features on the data supplied by the TempEval-2 task of SemEval-2010. Ex-perimental results show that the proposed method is benefit to Chinese time expression recognition.%提出了一种中文时间表达式识别方案.该方案在抽取传统特征的同时,根据中文时间表达式的特点,新加入语义角色( semantic roles,SR)特征构建特征向量,然后采用CRFs(条件随机场)进行识别.在SemEval2010评测的TempEval-2任务数据上进行实验,该方案识别中文时间表达式的F1(F1-measure)值达到85.6％,与未加入语义角色特征相比提高了5.2％.实验表明提出的方案在解决中文时间表达式识别问题上有较好的效果.