首页> 外文期刊>Information Processing & Management >AAEE - Automated evaluation of students' essays in Arabic language
【24h】

AAEE - Automated evaluation of students' essays in Arabic language

机译:AAEE-自动评估阿拉伯文学生的论文

获取原文
获取原文并翻译 | 示例

摘要

Assessing student's essay writing and providing thoughtful feedback is a truly labor-intensive and time-consuming task. With human instructors already overwhelmed, the alternate is to consider a computer-based grading. Recent advances have generated renewed interest in automatic evaluation of essays (AEE). The AEEs instantaneous feedback and more consistent grading helps students draft better essays. This work presents a system to automatically grade the school children essays in Arabic, calling it AAEE for "automatic Arabic essays evaluator". The system is modeled upon the scoring scheme followed by the school instructors in Saudi Arabia. The instructors had specific criteria upon which an essay is assessed. Putting these criteria together we developed a system that relies on Latent Semantic Analysis, and Rhetorical Structure Theory. With this design we are able to assess individual components of the essay such as language proficiency, structure of the essay etc. To test the system, we collected essays by local school children covering grades 7-12. A total of 350 different handwritten essays spanning eight different topics each transcribed into computer readable format. The AAEE shows that 90% of the test essays were correctly scored, and a correlation of 0.756 between automatic and teachers' scoring. This exceeds the human-human correlation of 0.709 for the Arabic essays.
机译:评估学生的论文写作并提供周到的反馈是一项真正的劳动密集型且耗时的任务。在人类导师已经不知所措的情况下,替代方法是考虑基于计算机的评分。最近的进展引起了人们对论文自动评估(AEE)的新兴趣。 AEE的即时反馈和更一致的评分有助于学生撰写更好的论文。这项工作提出了一种自动对阿拉伯文中小学生论文进行评分的系统,称其为“自动阿拉伯文论文评估者” AAEE。该系统以评分方案为模型,沙特阿拉伯的学校讲师遵循该评分方案。讲师有评估论文的特定标准。综合这些标准,我们开发了一个基于潜在语义分析和修辞结构理论的系统。通过这种设计,我们能够评估论文的各个组成部分,例如语言能力,论文结构等。为了测试系统,我们收集了当地7至12年级学生的论文。共有350篇不同的手写论文,涵盖了8个不同的主题,每个主题都被转录为计算机可读格式。 AAEE显示90%的测试论文均得到正确评分,自动评分与教师评分之间的相关系数为0.756。这超过了阿拉伯文中人与人的相关性0.709。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号