Predicting examinee performance based on a fuzzy cloud cognitive diagnosis framework in e-learning environment

The score profiles could be used to measure learners' skills proficiency via cognitive diagnosis models (CDMs) for predicting their performance in the future examination. The prediction results could provide important decision-making supports for personalized e-learning instruction. However, facing the possible complexity of skills, the uncertainty of learners' skill proficiency and the large-scale volume of score profiles, the existing CDMs have limitations in the measurement mechanisms and diagnostic efficiency. In this paper, we proposed an approach based on a fuzzy cloud cognitive diagnosis framework (FC-CDF) to predicting examinees' performance in e-learning environment. In this approach, the normal cloud models (NCMs) are utilized innovatively to measure the expectation, degree of variation and variation frequency of learners' skill proficiency, and each NCM is transformed into an interval fuzzy number to characterize the uncertainty of the skill proficiency for every learner. Combining the educational psychology hypothesis with the parameter estimation method, we could obtain the learners' skill proficiency level and the slip and guess factors relevant to each test item, based on which the learners' scores could be predicted in a future test. Finally, the experiments demonstrate that the proposed approach provides good accuracy and significantly reduces execution time for predicting examinee performance, compared with the existing methods.