AUTOMATED ARABIC ESSAY GRADING SYSTEM BASED ON F-SCORE AND ARABIC WORLDNET
Saeda A. Al Awaida,Bassam Al-Shargabi,Thamer Al-Rousan
Automated Essay Grading,Support Vector Machine,Arabic WordNet,Cosine Similarity,Natural Language Processing.
29-Jul.-2019 and 20-Aug.-2019
An Automated Essay Grading (AEG) system is designed to be used in universities, companies and schools, which
depends on Artificial Intelligence and Natural Language Processing technologies; as it has the capability to
improve the grading system in terms of overcoming cost, time and teacher effort while correcting the students’
essay questions and papers. The AEG system widespread use is due to its cost, accountability, standards and
technology; as that leads to the system being used and applied for multiple languages, such as English and French,
among others. On the other hand, limited research has been conducted to automate Arabic essay grading.
Therefore, this paper introduces an Arabic AEG system. In this paper, we propose a model for Arabic essay
grading based on F-score to extract features from student answers and model answers along with the use of the
Arabic WordNet (AWN) as a valuable knowledge-based method for semantic similarity. The purpose of using the
AWN is to find all related words from student answers to give the answer a student score. Students will not be
subject to injustice in terms of their marks in cases when they do not write the exact model answer, which
subsequently leads to an improvement of the Arabic AEG system to match human grading. The proposed model
is evaluated using Arabic essay dataset and the result shows that our proposed model produces a result which
matches human grading.