AUTOMATED ARABIC ESSAY SCORING BASED ON HYBRID STEMMING WITH WORDNET

Abstract

Schools, universities, and other educational institutions have been forced to close their doors because of the coronavirus outbreak. E-learning has become an option and has long been discussed about the need to integrate it into the educational process-learning uses a variety of evaluation methods, one of which is the essay. This research introduces a new model for Arabic Automated Essay Grading (AAEG) that has been developed to reduce human bias mistakes and costs while saving time. However, (AAEG) is still in its infancy. The model relies on new hybrid stemming with Arabic WordNet (AWN). The primary goal of stemming is reducing inflectional forms of words to root words. The hybrid method is based on different techniques: Extended Light Stemmer, ISRI, and looking at tables (AWN). Data used in this study consists of 3050 words with their roots were retrieved from (AWN) and then stemmed using algorithms (Light10, ISRI, Hybrid...). For evaluation, the metrics used were accuracy, precision, recall, and F1-score. While comparing the performance of the different stemming algorithms, the hybrid stemming method had the greatest results, therefore the (AAEG) will improve with Hybrid Stemming.