【作者】 石民； 陈小荷； 于丽丽； 李斌；

【Author】 SHI Min,CHEN Xiaohe,YU Lili,LI Bin Department of Linguistic Science and Technology,Nanjing Normal University,Nanjing 210097

【机构】 南京师范大学语言科学及技术系；

【摘要】 本文在计算机自然语言处理和古代汉语、特别是先秦文献的交叉领域进行了新的探索。首先对《左传》文本进行了词汇处理(分词和词性标注)和分析,然后采用条件随机场模型(CRF),基于两个模板进行自动分词、词性标注、分词标注一体化的对比实验。研究表明,一体化分词方法比单独分词的准确率和召回率均有明显提高,开放测试的最高F值达到了90.89%,满足古代汉语词汇研究和语料库建设的需求,而且较好地弥补了人工标注的不足。更多还原

【Abstract】 In this paper.we made a new exploration in the cross field between NLP and Ancient Chinese,in particular of the pre-Qin literature.We processed and analyzed the vocabularies of"Zuo Zhuan"text,then used the conditional random model(CRF) to do automatic WS,POS tagging,joint WS and POS tagging experiments based on two different templates. This research has shown that the P and R value of the integrated approach have markedly improved than the independent WS,and the highest F value has reached to 90.89%.This method meets the requirements of the study of Ancient Chinese vocabulary and corpus construction,and it can make up for the shortcomings of manual tagging.更多还原