Improving learning style identification by considering different weights of behavior patterns using particle swarm optimization

Matching the course content to students’ learning style has been shown to benefit students by improving their learning outcome, increasing satisfaction, and reducing the time needed to learn. Consequently, an accurate method for identifying these learning styles is of a high importance. Up to the present, there have been proposed several such methods that use students’ behavior in online courses to automatically identify their learning style. However, the precision of existing approaches peaks at approximately 80 %, thus leaving room for improvement. This paper introduces a novel approach, which combines the advantages of artificial/computational intelligence and rule-based techniques. More specifically, a rule-based method is extended to consider the different weights of behavior patterns using a particle swarm optimization algorithm. The approach has been evaluated with 75 students, and results show improved performance over similar state-of-the-art methods. By identifying learning styles with higher precision, students can benefit from adaptive courses that are tailored more precisely to their actual learning styles and teacher can benefit by being able to provide students with more helpful interventions.