Detecting Learning Style Based on Level of Knowledge

Abstract

One way to optimize online learning is by a personalized approach. One of the personalized models for online learning is knowing the learning styles of the learners. This is because learning styles that have been adopted in the early detection of learning can help learners follow a more optimal path in online learning. Currently the method used to detect learning styles uses a questionnaire, behaviour and datadriven. But the result of learning style detection of VARK by using all three methods of accuracy value is still below 70%. This is because the detection process does not take into account the prior knowledge side. Each learner must have different prior knowledge. Differences in prior knowledge can be the basis for detecting learning styles. There are three stages that are used to determine learning style, that is: construct prior knowledge, determination of level of knowledge and prediction of learning style. The stages of prior knowledge generation involve using online assessment and evaluation using the Weighted BLEU (WB) method. WB is a method that has the ability to detect learning answer. The learning answer will be indication level of prior knowledge. After prior knowledge is generated, the result of an assessment value determines the level of knowledge. Furthermore, the level of knowledge of each learner can be used to predict the VARK learning style of each learner. The learning style prediction process uses an artificial neural network. This research is able to achieve the result of learning style detection of VARK with an 83% accuracy level.