DEVELOPMENT OF AN INTELLIGENT SUBJECTIVE-BASED EXAMINATION MARKING AND SCORING SYSTEM USING MACHINE LEARNING TECHNIQUE

Subject Area: COMPUTER SCIENCE

This paper presents the development of an improved computer-based examination and automatic marking system using machine learning technique. The major aim is to develop a subjective examination system which allows lecturers to set theory-based questions on the online platform and then give students access to take the examination. It marks the questions answered after the examination and generate the scores instantly within the time boundary of the examination. The methodologies used in this paper are experimental research methodology. The Software Development Life Cycle (SDLC) employed was the water fall model, while the software design methodology adopted was also the water fall model. The methods are user registration, login, subjective exam setting, feature extraction, time control, answer matching and scoring system. The system was modeled with self-driving equations and Universal Modeling Language (UML). The model of the answer matching system was developed with K-NN algorithm while scoring system was modeled with rule-based approach. The matching algorithm was trained with classification application and then implemented with JavaScript and PHP for web-based Computer-Based Test (CBT) system alongside the rule-based scoring algorithm. The result of this system presented area under curve performance of 0.87 and accuracy of 87% when tested. The performance when compared with other CBT algorithms achieved better performance compared with the rest in terms of reliability, cost effectiveness and user friendliness. Keywords: Automatic Marking; Object-Oriented; Rapid application; CBT; UML; K-NN