基于BERT和题干要素语义增强的高考阅读理解自动答题  

AUTOMATIC ANSWER OF CHINESE COLLEGE ENTRANCE EXAMINATION CHINESE READING COMPREHENSION BASED ON BERT AND SEMANTIC ENHANCEMENT OF QUESTION ELEMENTS

作　　者：宋泽宇 王笑月 张虎[1] 李茹[1,2] Song Zeyu;Wang Xiaoyue;Zhang Hu;Li Ru(School of Computer and Information Technology,Shanxi University,Taiyuan 030006,Shanxi,China;Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education,Shanxi University,Taiyuan 030006,Shanxi,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：《计算机应用与软件》2023年第7期151-158,共8页Computer Applications and Software

基　　金：国家重点研发计划重点专项(2018YFB1005103);国家自然科学基金项目(61772324)。

摘　　要：高考阅读理解试题因其语言复杂度高和自动答题难度大,已成为机器阅读理解领域一项具有挑战性的任务。现有的答题方法普遍关注选项与材料的语义相似性,易于忽视题干信息对正确答案的要求,基于此,提出一种基于BERT与题干要素语义增强的高考阅读理解自动答题方法。通过构建问题模板的方式获取题干关键要素信息并生成问题标签;通过改写题干内容统一题干要求;将问题标签与BERT模型相结合完成答案选择。在高考数据集上的实验结果表明,该方法比多个典型的机器阅读理解基线模型取得了更好的效果。Because of its high language complexity and difficulty in answering questions automatically,the reading comprehension test of Chinese College Entrance Examination has become a challenging task in the field of machine reading comprehension.The existing answering methods generally pay attention to the semantic similarity between options and materials,and tend to ignore the requirement of correct answers for question.Based on this,this paper proposes an automatic answering method for the reading comprehension of Chinese College Entrance Examination based on BERT and semantic enhancement of question elements.The key element information of the question was obtained by constructing a question template,and the question label was generated according to the element information.By rewriting the content of the question,the requirements of the question were unified.The question label was combined with BERT model to complete the answer selection.The experimental results on the Chinese College Entrance Examination data set show that the proposed method achieves better results than several typical baseline models of machine reading comprehension.

关 键 词：高考阅读理解 选择题 题干关键要素信息 BERT 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]