Manuscript received December 10, 2023; revised December 25, 2023; accepted February 1, 2024; published June 17, 2024
Abstract—With the increased number of competitive examinees, adopting Multiple Choice Tests (MCTs) in most examinations has significantly shaped the assessment methodology. However, the success of this method depends on the quality of the items. Thus, selecting relevant items, balanced for difficulty and discrimination power, is crucial to guarantee the assessments’ validity and reliability. In this regard, integrating Artificial Intelligence (AI) provides promising prospects for further enhancing the item analysis and selection process. Therefore, this research aims to build a Machine- Learning (ML) model that discerns and selects items based on their difficulty and discrimination. This study employs the Artificial Neural Networks (ANN) method through binary classification models for item classification. The study’s experimental results demonstrate the proposed model’s efficacy, showcasing superior performance with an accuracy rate of 96% for item selection.
Keywords—e-assessment, competitive exams, items analysis, P-index, D-index, artificial intelligence, deep learning
Cite: Najoua Hrich, Mohamed Azekri, and Mohamed Khaldi, "Artificial Intelligence Item Analysis Tool for Educational Assessment: Case of Large-Scale Competitive Exams," International Journal of Information and Education Technology vol. 14, no. 6, pp. 822-827, 2024.