【作者】 彭洪保； 李茹； 段建勇；

【Author】 PENG Hongbao,LI Ru,DUAN Jianyong 1.School of Computer & Information Technology,Shanxi University,Taiyuan 030006,China; 2.Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education, Taiyuan Shanxi 030006,China 3.School of information engineering North China university of technology,Beijing 100144,China

【机构】 山西大学计算机与信息技术学院； 山西大学计算智能与中文信息处理教育部重点实验室； 北方工业大学信息工程学院；

【摘要】 语义角色标注是近些年来自然语言处理领域的一个新的研究热点。本文针对中文问句的自身语言特点,基于汉语框架网(Chinese FrameNet,CFN),提出了一种基于词性筛选和层叠条件随机场模型的汉语问句语义角色自动标注方法。该方法根据所需标注问句与语料库原有句子相似程度选择不同标注模型,并对2011条中文问句进行了自动标注实验。其中,框架识别准确率为86.7%,语义角色自动标注结果准确率为81.1%,召回率为74.9%。与D.Gildea等对英语框架元素语义角色自动标注结果准确率和召回率相比,取得了较好实验效果。更多还原

【Abstract】 Semantic Role Labeling is a new rising research in natural language processing field in recent years. According to the characteristics of Chinese questions,a method based on the selection of POS and Cascading conditional random field(CRF) model is proposed to research the automatic labeling of CFN Semantic Role for Chinese Questions.This method selects different labeling model in line with the similarity between the input question and the source sentences in corpus.The automatic labeling experiment for 2011 Chinese questions,in which shows that 86.7%precision for frame identification,81.1%precision for Automatic Labeling results and 74.9%recall respectively, which performs better than automatic labeling of Semantic Role for English Frame elements studied by D.Gildea.更多还原