北京邮电大学学报(社科版) ›› 2016, Vol. 18 ›› Issue (4): 108-116.

• 教育研究 • 上一篇    

主观题自动测评研究

  

  1. 北京邮电大学 自动化学院,北京100876
  • 收稿日期:2016-03-24 出版日期:2016-08-31

Automated Assessment of Subjective Tests

  1. School of Automation, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Received:2016-03-24 Online:2016-08-31

摘要:

主观题自动测评作为智能教育的关键技术之一是语言学和自然语言处理等相关学科交叉的一个研究方向,本文将主观题分为有参考答案和开放式答案两种类型,针对有参考答案的主观题依据语义相似度计算方式不同分为简单词形匹配、借助世界知识和依据大规模语料等三类,介绍计算机自动测评领域国内外在主观题方面的评测技术、方案的研究进展,总结各方案的优缺点提出借鉴深度学习的研究切入点,为汉语主观题测评研究提供方法和思路上的参考、启示。

关键词:  , 计算机自动批改, 主观题, 自动阅卷, 自然语言处理, 智能教育

Abstract:

As one of the key technologies of smart education, automated assessment of subjective tests (AAST) is a crossed research direction on linguistics and natural language processing (NLP) and related disciplines Subjective questions will be divided into two types which have reference answers and open answers, and then based on the ways computing semantic similarity, the subjective questions with reference answers will be divided into three types which are simple word shape matching, use the universal knowledge and based on large-scale corpus The advances in evaluation technology and program on the subjective questions in domestic and overseas are introduced Then the advantages and disadvantages of each way are summarized and research point of deep learning can be learned to provide reference and inspiration for AAST on Chinese

Key words: computer automatic correcting, subjective tests, automated assessment, natural language processing, smart education

中图分类号: