基于ChunkCRF的观点持有者识别与其在观点摘要中的应用.pdfVIP

2009 7 7 Journal of Chin s Comput r Sy st ms Vol. 30 No. 7 2009 ChunkCRF 1 2 1 宋　 锐 ,洪　莉 ,林鸿飞 1(, 116024) 2( , 122000) E-mail: hflin@ dlut . du. cn 　: 含有观点的文档中准确识别出观点的持有者是预处理步骤.通过建立 Chunk CRF模型对观点表达句 行观点持有者 的识别; 对于同一个观点句中含有多 个观点持有者的情况,借助语言学手段 行预处理,再利用模型 行观点持有者识别.在此 基础上还 行了观点的摘要与倾向性分析的工作.实验结果表明,基于Ch unkCRF的中文观点持有者识别方法达到了 80% 上以 的准确率,并且能够更好的配合观点的摘要与倾向性分析工作. : ChunkCRF;;; : 39 1. 3　　　　　: 　　　　　　 : 1000-1220( 2009) 07-1462-05 TP A Chunk-CRF-based Opinion Holder Identification and Application to Opinion Summarization 1 2 1 SONG Rui , HONG Li , LIN Hong-f i 1(Department of Computer Science and Engineering , Dalian University of Technology , Dalian 116024, China) 2( ’ , 122000, ) Chaoyang Teacher s College Chaoyang China : - - Abstract Accurat opinion hold r id ntification is an important pr proc ssing for opinion summarization bas d on opinion hold r. In this pap r, a Chunk CRF mod l w as construct d to divid th opinionat d s nt nc into particular chunks with th ; , - - , aim of ff ctiv ly id ntify th opinion sourc s at th sam tim in multi Opinion Hold r cas s syntactic analy sis was mad us of b for applying CRF mod l. Mor ov r , opinion summarization and polarity analy sis w r also giv n. Exp rim nt r sults - 80% , , show d th ChunkCRF bas d m thod id ntifi d opinion hold rs with pr cision ov r mor ov r it could also assist th opinion summarization and polarity analy sis w ll. : ; ; ; Key words ch unkCRF opinion h old r id ntification opinion summarization opinion polarity analysis 1　　 , ,