面向中文专利文献的单层并列结构识别

Identification of Non-nest Coordination Base on Chinese Patent Literature

结合条件随机场与伯克利句法分析器对中文专利文献中的单层并列结构进行识别。在经过分词和词性标注的中文专利语料的基础上,分别运用条件随机场和伯克利句法分析器对专利语料中的单层并列结构进行识别,提取两种机器模型相同的和识别结果中满足相应规则的并列结构识别结果。实验结果表明,该方法有效的识别了专利文献中的单层并列结构,取得了73.09%的F值。

Identification of non-nest coordination on Chinese patent corpus can combine with conditional random fields(CRF)and the Berkeley parser Based Chinese word Segment and Tagging,Identification of non-nest coordination by using conditional random fields and the Berkeley parser Pick up the identified result of the two same machine models and satisfy the relevant rules' coordination result on it.The experimental results show that this approach could identify the non-nest coordination on the patent literature effectively and get the F value of 73.09%.