In e-learning, recommendation systems have proven to be highly efficient for improving learners' performance and knowledge. They can manage the different pedagogical resources and simplify the workload for the instructor and learners as well. Throughout the years, recommendation systems in e-learning have wit-nessed a major evolution since the 2000s. Several aspects have been developed, including techniques involved, test data (...). In this respect, this paper analyses the evolution of recommendation systems in e-learning since 2000 with a focus on the evolution sides. It furthermore addresses areas not fully addressed to date. A set of recommendation systems is identified and then analysed in order to define techniques used, as well as algorithms deployed.

Generative Artificial Intelligence (GAI), also known as creative AI, is a technology capable of independently producing original and creative content, such as text, images, videos, music, or code. It uses machine learning algorithms to analyze and learn from a vast amount of data to generate similar or even innovative content. This type of AI is used in several fields, such as health for the production of new drugs, design to generate adaptive designs, video games to create immersive environments, and artificial intelligence to generate training data for new networks. Seizing the opportunities offered by AI innovations in education is crucial. We cannot afford to let time pass without doing anything and without benefiting from this technique. It is time to enhance traditional learning methods and adopt more innovative approaches that take advantage of the potential of AI, like content and exam generation, and also improve students' learning experience by exploiting their closeness to the technology. This paper aims to answer the question of when and how to use artificial intelligence responsibly, considering it as a technology and limiting its excessive use. Additionally, we present the impact of GAI through a comparative study of student performance in three exam types: a classical MCQ exam written by the teacher, an adaptive MCQ exam generated by GAI from a topic, and a third MCQ exam generated by GAI based on the content of the provided documents. This study is conducted on students undergoing a software testing training program. By evaluating their performance in the AI-generated adaptive MCQ exams and the traditional MCQ exam, we can gain insights into the potential of this technology to enhance educational practices and provide personalized learning experiences. This research will contribute to the ongoing discussion on responsible and effective utilization of GAI in education, paving the way for future advancements in the field.

The use of rail transport is increasing. Damaged rails interrupt traffic to carry out repairs. In fact, when a conductor or railway operator reports a damaged rail that interrupted the traffic in the affected area. A team of specialized agents dispatch to the site and carries out the repairs. Hence, the importance of automation of railway track faults detection to ensure track safety and reduce maintenance costs. In this work, we propose a method using image processing technologies and deep learning networks. We have studied the correlation effects of MobileNetV2 and optimization algorithms on accuracy and other performance metrics to generate a model that can achieve good performance in classifying railway track faults. The results show that the Rmsprop can improve the effectiveness of feature extraction and classification of MobileNetV2.

In distance learning environments, learner engagement directly impacts attention, motivation, and academic performance. Signs of fatigue, negative affect, or critical remarks can warn of growing disengagement and potential dropout. However, most existing approaches rely on a single modality, visual or text-based, without providing a general view of learners’ cognitive and affective states. We propose a multimodal system that integrates three complementary analyzes: (1) a CNN-LSTM model augmented with warning signs such as PERCLOS and yawning frequency for fatigue detection, (2) facial emotion recognition by EmoNet and an LSTM to handle temporal dynamics, and (3) sentiment analysis of feedback by a fine-tuned BERT model. It was evaluated on three public benchmarks: DAiSEE for fatigue, AffectNet for emotion, and MOOC Review (Coursera) for sentiment analysis. The results show a precision of 88.5% for fatigue detection, 70% for emotion detection, and 91.5% for sentiment analysis. Aggregating these cues enables an accurate identification of disengagement periods and triggers individualized pedagogical interventions. These results, although based on independently sourced datasets, demonstrate the feasibility of an integrated approach to detecting disengagement and open the door to emotionally intelligent learning systems with potential for future work in real-time content personalization and adaptive learning assistance.

This article aims to present a new method of evaluating learners by communities on Facebook groups which based on their interactions. The objective of our study is to set up a community learning structure according to the learners' levels. In this context, we have proposed a new algorithm to detect and evaluate learning communities. Our algorithm consists of two phases. The first phase aims to evaluate learners by measuring their degrees of ‘Safely’. The second phase is used to detect communities. These two phases will be repeated until the best community structure is found. Finally, we test the performance of our proposed approach on five Facebook groups. Our algorithm gives good results compared to other community detection algorithms.