基于浅层语义树核的阅读理解答案句抽取

被引:13
作者
张志昌
张宇
刘挺
李生
机构
[1] 哈尔滨工业大学计算机学院信息检索研究室
关键词
计算机应用; 中文信息处理; 阅读理解; 答案句抽取; 浅层语义; 树核;
D O I
暂无
中图分类号
TP391.1 [文字信息处理];
学科分类号
081203 ; 0835 ;
摘要
阅读理解系统是通过对一篇自然语言文本的分析理解,对用户根据该文本所提的问题,自动抽取或者生成答案。本文提出一种利用浅层语义信息的英文阅读理解抽取方法,首先将问题和所有候选句的语义角色标注结果表示成树状结构,用树核(tree kernel)的方法计算问题和每个候选句之间的语义结构相似度,将该相似度值和词袋方法获得的词匹配数融合在一起,选择具有最高分值的候选句作为最终的答案句。在Remedia测试语料上,本文方法取得43.3%的HumSent准确率。
引用
收藏
页码:80 / 86
页数:7
相关论文
共 9 条
[1]  
A MaximumEntropy Framework that Integrates WordDependencies and Grammatical Relations for ReadingComprehension. Kui Xu,Helen Meng,Fuliang Weng. Proceedings of the HumanLanguage Technology Conference of the NAACL . 2006
[2]  
A Rule-basedQuestion Answering System for ReadingComprehension Tests. Ellen Riloff,and Michael Thelen. Proceedings of ANLP/NAACL 2000 Workshop on Reading ComprehensionTests as Evaluation for Computer-Based LanguageUnderstanding Systems . 2000
[3]  
Using Verb DependencyMatching in a Reading Comprehension System. Kui Xu,and Helen Meng. First Asia Information Retrieval Symposium(AIRS2004) . 2004
[4]  
Reading Comprehension Programs in aStatistical-Language-Processing Class. Eugene Charniak,Yasemin Altun,Rodrigo de SalvoBraz. Proceedings of the ANLP/NAACL 2000 Workshop onReading Comprehension Tests as Evaluation forComputer-Based Language Understanding System . 2000
[5]  
Exploiting Redundancy in QuestionAnswering. Charles L.A.Clarke,Gordon V.Cormack,ThomasR.Lynam. Proceedings of the 24thACMConference on Research and Development inInformation Retrieval(SIGIR-2001) . 2001
[6]  
WordNet:an On-line lexicaldatabase. George A.Miller. International Journal of Lexicography . 1990
[7]  
Deep Read:A ReadingComprehension System. Lynette Hirschman,Marc Light,Eric Breck,andJohn D.Burger. Proceedings of the37thAnnual Meeting of the Association forComputational Linguistics . 1999
[8]  
New rankingalgorithms for parsing and tagging:Kernels overdiscrete structures,and the voted perception. Michael Collins,and Nigel Duffy. Proceedings of the 40th Annual Meeting onAssociation for Computational Linguistics(ACL’02) . 2002
[9]  
A Machine Learning Approach to AnsweringQuestions for Reading Comprehension Tests. Hwee Tou Ng,Leong Hwee Teo,Jennifer Lai PhengKwan. Proceedings of the 2000 Joint SIGDAT Conference onEmpirical Methods in Natural Language Processingand Very Large Corpora . 2000