A new approach for learners' modeling in e-learning environment using LMS logs analysis

Nowadays due to the necessity to develop personalized e-learning, knowledge about learners is required. A questionnaire is a practical way to gather information on learning style. There are some problems in questionnaire usage such as reluctance to answer questions, random guesses, taking too much time and invalid answers. In this paper, we have introduced the new parameters saved in Learning Management System(LMS) log files that are equivalent to learning approaches questionnaire. This research has been performed using40MSce-learning students on four different courses and in three phases(stages)at Tarbiat Modares University. The results derived from the questionnaire were acquired according to the learners' learning style in the first phase and in the second one were derived on Bayesian Network(BN) approach. The results based on log file and network help were taken in the third phase. It is mentioned that the third results had the least uncertainty range due to BN approach. Finally equivalent parameters for each question and for each learning style dimension are introduced based on some applied LMS package such as Moodle and Sakai. The main points in this research are as below: learning style extraction based on all the learners' activities in LMS (in a day, in a course and in a semester) and not only based on his/her related activities. For instance, the results show that the learner is more active and general if he/she considers his/her other class mates's profile or add/complete (in forums) during exact time to study. These results can be applied in Personalized Intelligent Tutoring System(PITS) in e-learning environment and so in designing recommender system to increase learners' and tutors' satisfaction level.