Trace Analysis of Educational Videos to Identify the Groups of Learners with Similar Profiles Amel Behaz 1 , Hamouma Moumen 1* , Wafa Nouari 2 1 Computer Science Department, University of Batna 2, Batna 05000, Algeria 2 English Department, University of Batna 2, Batna 05000, Algeria Corresponding Author Email: hamouma.moumen@univ-batna2.dz https://doi.org/10.18280/isi.280110 ABSTRACT Received: 14 October 2022 Accepted: 20 January 2023 The involvement of teaching videos increases the learners’ psychological stimulation during the learning process. In fact, this type of resource can facilitate comprehension by approaching the content at the learners own pace and coming back to it as many times as necessary. In this particular case, the possibility of keeping a trace of video playing will be considered for the teacher an important asset in monitoring the learning progress of the learners. The objective of our work is to analyze the learner traces when he/she is playing a video. We propose usable restitutions to improve the online learning. Among these restitutions, we suggest to the teacher groups of learners who will be able to work together since they have the same video playing strategies. Throughout this analysis, the results are from a specific case study on which we applied unsupervised classification methods to identify groups of learners with similar profiles. Keywords: learning analytics, learner profile, educational data mining, clustering trace analysis 1. INTRODUCTION Digital technologies continue to display an increasingly important role in education: digital textbooks, learning games, e-learning, MOOC, etc. These new Digital technologies can help us make teaching engaging and creative. They have become a necessity to guarantee learning. For example, during the COVID-19 pandemic, countries that lacked digital technology infrastructure experienced disruptions and learning losses. Thus, the use of these technologies offers a potential on the performance of learners. Today, in addition to the increased learning needs, the possibility of distance learning is a decisive advantage [1]. Learning analytics can improve learning practice by transforming the ways we support into learning processes. Historically, learning analytics has emerged as a promising area of research that extracts useful information from educational databases to understand students’ progress and performance. Today, the Web is gradually replacing the amphitheater and confidently teaching online video courses with entertainment. Thus, such success has led to a pedagogical revolution since learners are no longer rushing to their lecture halls yet eager to watch online videos courses that eventually pose new pedagogical challenges. Actually, this pedagogical strategy can pave the way for the teacher to keep a trace of video playing in monitoring the learners’ learning progress. For example, a teacher can use it to personalize a learning path. A learner can also visualize his/her learning line and positioning in relation to other learners. Also, a researcher will be able to develop these traces in order to produce new knowledge and provide feedback relevant or diagnostic to assist the teacher in the course. Today, learning institutions seek ways to collect, manage, analyze and exploit data from learners for the facilitation of learning processes. There are two types of data learners: qualitative data which correspond to direct answers, forms, or other traces which correspond to all interactions data with the learning environment. Among the traces we can cite the consultation time of course pages, logs on pages, quiz attempts and score. We are interested exactly in the interactions’ traces playing the video courses by the learners. This information will allow teachers to detect difficulties of assimilation and monitor the behavior of their learners. However, it is a challenge that we must take up to optimize the lessons as well as possible and improve the learners’ paths while learning. Teachers need tools to improve monitoring learners and understanding their behaviors. The tools collect information from learners and tutors (teachers) and use it in their programs to track learning progress. Therefore, these tools must allow clear and widely tested and accessible visualizations. Analytical data from monitoring of reading of resources can be used to guide learners’ progress towards the goals set: Monitoring and analysis, Prediction and intervention, Tutoring, and mentoring (coaching), Monitoring and feedback, Adaptation, Personalization, recommendation, and Reflection [2]. We have noticed some disengagement among learners during distance learning activities. Therefore, the implementation of a new pedagogical approach is necessary to improve the involvement of these learners during the learning process. For example, we suggest groups of learners who can work together in practical activities. This from the analysis of the traces of the learners during an activity. In this paper, our goal is to study and improve the way learner behavioral profiles are return to teachers. We will keep in graphs traces of pedagogical online video courses, and through these graph’s data we apply two unsupervised classification methods to identify groups of learners with similar profiles. Ingénierie des Systèmes d’Information Vol. 28, No. 1, February, 2023, pp. 97-104 Journal homepage: http://iieta.org/journals/isi 97