Scholarlink

In traditional essay grading, the challenges are the diversity of responses to essay questions and the large number of students. In addition, transparency regarding the value results is immediately required to document student evaluations. This paper presents a new online essay grading model developed for the Open University of Indonesia (Universitas Terbuka Indonesia), which combines the classification capabilities of Gaussian Naive Bayes, TF-IDF, and quality features. The algorithm generated the probabilistic model of scoring classes based on different student answering styles, such as grammar, coherence, and argumentation. The proposed model is tested on a large dataset of student answer essays and applied to eight subjects. The model reduces teachers’ grading workload and provides timely and constructive feedback to students. The model also shows the improvement of a more effective and productive learning environment and provides a scalable solution to the challenges of assessing essays in higher education with an accuracy rate of 97.04%.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Online Essay Scoring Model Based on a Combination of Gaussian Naive Bayes, TF-IDF, and Quality Features for Open University

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Azhari Azhari,
  • Agus Santoso,
  • Anak Agung Putri Ratna,
  • Reza Pulungan,
  • Hayyu Ilham Wicaksono,
  • Ranggajaya Ciptawan,
  • Uffi Nadzima,
  • Aszani,
  • Jasson Prestiliano

摘要

In traditional essay grading, the challenges are the diversity of responses to essay questions and the large number of students. In addition, transparency regarding the value results is immediately required to document student evaluations. This paper presents a new online essay grading model developed for the Open University of Indonesia (Universitas Terbuka Indonesia), which combines the classification capabilities of Gaussian Naive Bayes, TF-IDF, and quality features. The algorithm generated the probabilistic model of scoring classes based on different student answering styles, such as grammar, coherence, and argumentation. The proposed model is tested on a large dataset of student answer essays and applied to eight subjects. The model reduces teachers’ grading workload and provides timely and constructive feedback to students. The model also shows the improvement of a more effective and productive learning environment and provides a scalable solution to the challenges of assessing essays in higher education with an accuracy rate of 97.04%.