面向高考阅读理解观点类问题的答案抽取方法  

Answer Extraction of Opinions Question in Reading Comprehension of Prose for College Entrance Examination

作　　者：王素格[1,2] 李书鸣 陈鑫[1] 穆婉青 乔霈 WANG Suge;LI Shuming;CHEN Xin;MU Wanqing;QIAO Pei(School of Computer and Information Technology,Shanxi University,Taiyuan 030006,China;Key Laboratory of Ministry of Education for Computation Intelligence and Chinese Information Processing, Shanxi University,Taiyuan 030006,China)

机构地区：[1]山西大学计算机与信息技术学院,山西太原030006 [2]山西大学计算智能与中文信息处理教育部重点实验室,山西太原030006

出　　处：《郑州大学学报（理学版）》2018年第1期54-59,共6页Journal of Zhengzhou University:Natural Science Edition

基　　金：国家"863"高技术项目(2015AA015407);国家自然科学基金项目(61573231;61632011;61672331)

摘　　要：为了克服高考阅读理解观点类问题中仅仅利用关键词进行答案句抽取的不足,通过对单篇文章建立LDA(latent dirichlet allocation)模型,计算阅读材料中每一句话与问题的主题分布相似度.利用SVM(support vector machine)分类器,对主题分布相似度高的句子进行分类,用于获取具有观点特征的候选观点句,最后将其与问题观点进行相似度计算并排序,选取排名前α个句子作为题目的答案句.该方法在标注好的高考观点类问题的答案句识别中,总体正确率达到了47.8%,并能将答案句的排序提前.实验结果表明,散文阅读理解题中采用多主题方法比单主题时具有更好的效果.A new method was proposed to extract answers in reading comprehension to overcome shortcomings of key-words.First,through the LDA(latent dirichlet allocation)model,the similarity between each sentence of the article and the topics of the question were calculated.Then,the sentences with high similarity were classified.Finally,the similarity between the sentences in article and the question were calculated and sorted to choose the firstαsentence as the answer.Experimental results showed that the proposed method could achieve an accuracy of 47.8%in the college entrance examination corpus overall,and could sort the answer sentences to the front.The multi-topics approach had a better effect than a single-topic model for the answer extraction of the prose reading comprehension.

关 键 词：问答系统 语义相似 阅读理解 主题 观点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]