Design of online examination evaluation model for college students based on fuzzy reasoning system

Abstract

Evaluating online assessments with accuracy and efficacy presents difficulties due to inconsistencies in student responses, varying degrees of precision, and the complexity of the questions. Conventional methods, typically binary, sometimes fail to encapsulate these intricacies, resulting in inequitable or inconsistent outcomes. A model for online examination evaluation employing a fuzzy reasoning system (OEEM-FRS) is presented to enhance fairness, adaptability, and coherence in assessing student performance. The framework utilizes adaptive membership functions to transform correctness, completeness, and completion time into language variables, with Mamdani-type inference facilitating explicit decision-making. A Genetic Algorithm is used to enhance membership borders and rule weights, consequently augmenting fairness, adaptability, and stability. Additionally, a supervised machine learning fusion module improves predictive accuracy by correlating fuzzy scores with real data distributions. The experimental assessment of student performance datasets reveals substantial enhancements, demonstrating an accuracy of 94.6%, a fairness measure of 0.93, and a consistency score of 0.91, relative to leading approaches. The results underscore that OEEM-FRS can provide evaluations that are comprehensible, flexible, and robust. This paper presents a scalable and pragmatic framework for online evaluation systems, facilitating equitable decision-making in contemporary educational environments.