Natural Language Processing Skills of Learnings Through the Use of Conceptual Interpretation and Textual Responses

Main Article Content

Ramlan Mustapha, Sti Habsah Sh Zahari, Ahmad Zam Hariro, A. Jailani Che Abas, Norsafizar Mohd Noor, Nor Adila Mohd Noor, Senin MS, Zaharuddin Ibrahim

Abstract

The goal of this study was to see how natural language processing (NLP) approaches could be used in an educational setting to assess Learners' conceptual knowledge based on their short answer responses. Completion testing stimulates and offers response on Learners' abstract knowledge, which is frequently overlooked in automated grading. Automated formative assessment, which provides insights into conceptual comprehension as needed, benefits both Learners and instructors, especially in online education and large cohorts. It employed the ELECTRO-small, Roberto-base, XXLNET-base, and ALBERTO-V3 NLP machine learning models. These two parts of data shed light on Learners' conceptual understanding as well as the nature of their comprehension. It used high-performance NLP models to build a free-text validity ensemble with accuracies ranging from 91.46 percent to 98.66 percent for judging the validity of Learners' justifications. With precisions ranging from 93.07 percent to 99.46 percent, it suggested a generic, non-question-specific Response model for categorizing responses as high or low confidence.  Because of the great presentation of these models and their adaptability to lesser data sets, instructors have an exclusive chance to incorporate these approaches into their lectures.

Article Details

Section
Articles