Understanding and predicting learner performance is essential for designing effective educational strategies and supporting students who need additional help. Although graph neural networks (GNNs) show strong potential for modeling learner–content interactions, many knowledge-based GNN approaches struggle to generalize across different educational contexts due to their reliance on predefined knowledge structures. This study introduces an independent GNN framework based on GraphSAGE, evaluated on five real-world datasets from diverse learning environments. The model captures learning interactions without depending on knowledge-tracing assumptions and achieves strong performance (AUC = 0.87). Using inductive learning, GraphSAGE generates embeddings for new learners and items without retraining, improving adaptability and scalability. Consistent results across datasets show minimal need for domain-specific tuning, highlighting the model's potential for deployment in various learning systems to support adaptive pathways, early interventions, and equitable resource allocation.

Knowledge Tracing (KT) is a fundamental task in intelligent education systems, designed to track students' evolving knowledge states and predict their future performance. While Deep Learning-based Knowledge Tracing (DLKT) models have advanced the field, they often face significant limitations in jointly capturing short-term performance fluctuations and long-term knowledge retention, which restricts their predictive precision in complex learning trajectories. This paper proposes the Extended Deep Knowledge Tracing (xDKT) model, which integrates the Extended Long Short-Term Memory (xLSTM) architecture to enhance multi-scale temporal learning representations. Specifically, through rigorous ablation studies over extended learning sequences (up to 1000 steps), our analysis indicates that the exponential gating and advanced scalar memory of sLSTM units are the primary drivers of performance. This architecture effectively captures both short-term performance shifts and long-term knowledge retention without the vanishing gradient degradation inherent to standard LSTMs. We evaluate xDKT across six diverse benchmark datasets, including Synthetic, Algebra2005-2006, Statics2011, and the ASSISTments series, covering over 22,000 learners. Experimental results show that xDKT yields improved Area Under the ROC Curve (AUC) scores on Statics2011 (0.8562) and ASSISTments2009 (0.8318) compared to baseline models such as DKT, DKVMN, and AKT. Finally, through extensive validation, these findings suggest that xDKT architecture provides a robust and promising framework for accurate and adaptive learning environments.

Multi-tier web applications face significant implementation challenges in DevSecOps, including tool integration complexity, automation gaps, and cultural resistance. This study presents a proxy-based DevSecOps framework grounded in a formal architectural decomposition that transforms classical O (n & times; m) pipeline integration complexity into additive O(n(1 )& times; m(1)) + O(n(2) & times; m(2)) components through the separation of static and dynamic execution contexts. The framework is instantiated for PHP-based multi-tier web applications deployed on AWS infrastructure using Terraform-managed Infrastructure as Code principles; all ecosystem coherence metrics, toolset analysis, and complexity computations are derived within this specific technology context, and generalization to other language ecosystems or cloud platforms constitutes a boundary condition discussed in Section V. Theoretical contributions include: 1) a formal proxy pipeline architecture with mathematical complexity analysis demonstrating that complexity reduction is guaranteed when tool ecosystem coherence exceeds 70%; 2) systematic tool integration using PHP-specific tooling, yielding a theoretical ecosystem coherence of 76.9%; and 3) theoretical validation addressing 18 out of 20 identified DevSecOps implementation challenges. Mathematical analysis theoretically predicts a 48.13% reduction in tool integration conflicts and a 61.9% toolset reduction relative to traditional monolithic pipelines through context separation. All quantitative figures presented above are theoretically derived predictions, not empirically measured outcomes. They are formalized as falsifiable hypotheses H1 through H5 in Section V, with empirical validation identified as the primary direction for future work.

Intelligent tutoring systems generate a large volume of data, which becomes particularly valuable when effectively leveraged for learner performance prediction in adaptive learning environments. In this context, the speed and predictive accuracy of machine learning models are crucial, as they determine the system's ability to deliver timely and relevantinsights and support responsive, personalized instruction. Enhancing model speed not only increases tutoring efficiency but also improves the adaptability of educational systems to learners' needs. This study introduces an approach aimed at improving the execution time of three logistic regression-based models widely used for learner performance prediction: DAS3H (Item Difficulty, Student Ability, Skill, and Student Skill Practice History), AFM (Additive Factor Model), and PFA (Performance Factor Analysis). The proposed optimization reduces the complexity of the Q-matrix that links each item to its required knowledge components by simplifying its structure while preserving pedagogical relevance. An empirical evaluation was conducted on four real-world datasets collected from online tutoring platforms. The results demonstrate that the proposed approach, called Fast E-learning Recommendation (FER), significantly improves the execution speed of the three models while maintaining comparable predictive performance across datasets.

Learner performance prediction involves analyzing historical learner interactions within educational systems to forecast future success. Educational data mining models are increasingly used to personalize learning experiences. Recently, Graph Neural Networks have emerged as a powerful tool for modeling complex relationships in academic data. However, many state-of-the-art models are limited to single learning environments, which constrains their generalizability across diverse educational contexts. Moreover, these models often rely on hybrid architectures that integrate graph-based approaches with domain-specific frameworks-such as knowledge tracing-thereby introducing dependencies on traditional methods and limiting scalability. This study evaluates the effectiveness of several graph neural networks architectures, including Graph Convolutional Networks, Graph Attention Networks, and GraphSAGE, across multiple educational datasets: Assistments17, Assistments09, Statics, Spanish, and Moodle-Morocco. A knowledge graph was constructed using edge classification and representation learning, with Graph Neural Networks embeddings subsequently integrated into ensemble models such as XGBoost and LightGBM. The results show that Graph Neural Networks deliver strong baseline performance and generalize well across datasets, achieving an AUC of 0.87. Furthermore, the integration of ensemble models improves key performance metrics such as recall and F1-score, particularly for imbalanced datasets, reaching an AUC of 0.93. These findings highlight the potential of combining Graph Neural Networks with ensemble methods to create scalable and generalizable learner performance prediction systems.

The huge amount of data generated by an Intelligent Tutoring System becomes useful when analyzed in an appropriate way to provide significant insights about learners, especially his or her performance. Performance data retrieved from historical interactions is the main engine for learner performance prediction, where the likelihood of the learner answering correctly future questions is calculated. Modeling learner performance can provide significant insights into individual students to promote successful learning and maximize educational achievement. This study aims to enhance the learner performance prediction of some logistic regression-based models, namely Item Response Theory, Performance Factor Analysis, and DAS3H using XGBoost, including an empirical comparison of eight real-world datasets, containing performance log data collected from different online intelligent tutoring systems, involving the first time a new dataset from Moodle Morocco. The results have demonstrated that the XGBoost has enhanced PFA predictive performance on seven datasets with an AUC of up 0.88 and improved the DAS3H AUC on the ASSISTment17 dataset while conserving almost the same predictive results for Item Response Theory on some datasets.

The education sector has never been so shaken up as much as this past year. COVID-19 has imposed new rules. Several countries were forced to switch overnight from a traditional educational model to a full eLearning one. Like most other countries, the Moroccan government decided to promote distance learning by implementing several initiatives, though they remained at an embryonic stage. To contribute to the movement of transforming the national educational landscape, we aimed to develop a solution that will leverage the technological advances in this field and influence the ways students learn. This will be possible by providing learners with the latest features enabling online and adaptive learning modes. Hence, the purpose of this first study is to provide an empirical evaluation of the existing open source Ed-tech projects, which will serve as the basis for the development of our global adaptive eLearning solution. Unlike existing work, which is based on literature reviews to compare the existing adaptive eLearning platforms, we have used the OpenBRR assessment methodology as a comparison methodology due to its flexibility and ease of use. This work will help us to understand the concepts of adaptivity in education. It will also describe the most popular open source Maturity Models as well as provide a clear idea about the differences between these Ed-tech open source solutions.

Students' Performance prediction has attracted the interest of many researchers. Predictive models that predict whether a learner will complete a task successfully or not have several applications. Such models are used by Intelligent Tutoring Systems (ITS) to determine students' skills. The measurement of students' performance during their progress can provide critical information to students, teachers, and administrators. In this paper, we are going to evaluate and compare three models PFA, IRT and BKT on three datasets: ASSISTment, Cognitive Tutor, and a set of data extracted from learners' traces when they interact with a MOOC platform of a Moroccan university computer science course.

The distance education has become an indispensable teaching method, especially after the situation of Covid-19. For this reason, educational institutions are looking for e-learning platforms that offer better course management, ease of administration, user friendly, and achieving the learning objectives. Adaptive learning is considered an active research area, it enables to detect learning style of learners based on their behaviors and learning purposes in order to recommend relevant course materials. The objective of this article is to present an overview of personalization in the traditional learning system and the new developed systems as well as the approaches used to understand the learner’s individual needs. Furthermore, this work analyzes the problems in these systems and presents the prospect of development.

The traditional Internet architecture provides applications with a stable end-to-end connection between a requester and a source node holding the information. However, in a vehicular environment, this connection is not stable due to the mobility of vehicles that results in location change, from the requester, the source, or both. To overcome these situations, Named Data Networking (NDN) has been proposed as new architecture for data retrieval and mobility support. This new paradigm enhances content access and dissemination by decoupling the content from its original location. In this paper, we introduce the forwarding in NDN-based VANETs and highlight its benefits and limitations. We propose a classification of NDN-based VANETs forwarding strategies, then detail the representative schemes. Furthermore, we provide a review and comparison of existing forwarding strategies in terms of various attributes such as transmission mode, forwarding strategies, changes of NDN architecture, application scenarios, problem addressed, evaluation metrics, simulation platform. Finally, we conclude our contribution by identifying main open research challenges that can be exploited for future works. We believe that this survey will help NDN-based VANETs researchers community to easily understand the forwarding in vehicular environment, overcome some repeated solutions and gives further inspiration to design a new protocols to improve NDN-based VANETs networks in this context for further relevant research works.

In the era of big data, recommender systems (RSs) have become growing essential tools. They represent important machine learning solutions that mainly contribute to keeping users engaged with personalized content in e-platforms. Several RSs have been proposed in the literature, and most of them have focused on English content. However, for content in other languages like Arabic, very restricted works have been done to develop RSs. In recent times, the Arabic content on the Web has increased significantly because of the growing number of Arabic web users. This highlights the need for building RSs for Arabic content. To better handle this challenge, we decided to provide the research community with a novel deep learning (DL)-based RS devoted to Arabic content. The main goal of the proposed RS is to predict user preferences from textual reviews written in the Arabic language. This is achieved by combining two independent DL techniques into one system: a convolutional neural network (CNN)-text processor for representing users and items; and a neural network, in particular, a multi-layer perceptron (MLP) to estimate interactions between user-item pairs. Extensive experiments on four large-scale Arabic datasets demonstrate that our proposed system can achieve better prediction accuracy than other state-of-the-art alternatives. Notably, it improves the MSE between 0.84% and 16.96%, and the MAE between 0.14% and 13.71%. This work is the first attempt designed to deal with a large volume of data in the Arabic context, opening up new research possibilities for future developments of Arabic RSs.

With the upsurge of data traffic due to the change in customer behavior towards the use of telecommunications services, fostered by the current global health situation (mainly due to Covid-19), the telecommunications operators have a golden opportunity to create new sources of revenues using Big Data Analytics (BDA) solutions. Looking to setting up a BDA project, we faced several challenges, notably, in terms of choice of the technical solution from the plethora of the existing tools, and the choice of the governance methodologies for governing the project and the data. The majority of research documents related to the telecommunications industry have not addressed BDA project implementation from start to finish. The purpose of this study focuses on a BDA telecommunications project, namely, Project’s Governance, Architecture, Data Governance and the BDA Project’s Team. The last part of this study presents useful BDA use cases, in terms of applications enabling revenue creation and cost optimization. It appears that this work will facilitate the implementation of BDA projects, and enable telecommunications operators to have a better understanding about the fundamental aspects to be focused on. It is therefore, a study that will contribute positively toward such goal.

Recommender systems (RSs) have been used worldwide in several fields to facilitate the tourists’ planning activities. Tourism is one of the fields that uses RSs to reduce the overload of information to end users. Accordingly, tourists get recommendations that are most suitable to their profiles. This paper presents a detailed overview of the tourism recommender systems that were developed since 2008. It focuses mainly on the content based systems and their applications in the tourism field.

Because of the sanitary crisis created by the appearance of covid-19, the education system was forced to change its system to a flexible online one; the adaptive e-learning has become the pedagogical model of the first choice for successful education in the era of 2021. Adaptive e-learning platforms exist in abundance to offer personalized knowledge to improve learners’ personal skills, respond to their needs, and satisfy their interests, allowing them to master their learning. The core purpose of this paper is to provide the current state of the art by reviewing the existing adaptive, fee-based learning platforms based on a survey that was conducted to define the characteristics and features of each platform, with the vision to help learners choose The most suitable adaptive e-learning platform.

Detecting outliers in real-time is increasingly important for many real-world applications such as detecting abnormal heart activity, intrusions to systems, spams or abnormal credit card transactions. However, detecting outliers in data streams rises many challenges such as high-dimensionality, dynamic data distribution and unpredictable relationships. Our simulations demonstrate that some advanced solutions still show drawbacks. In this paper, first, we improve the capacity to detect outliers of both micro-clusters based algorithms (MCOD) and distance-based algorithms (Abstract-C and Exact-Storm) known for their performance. This is by adding a layer called LiCS that classifies online the K-nearest-neighbors (Knn) of each node based on their evolutionary status. This layer aggregates the results and uses a count threshold to better classify nodes. Experiments on SpamBase datasets confirmed that our technique enhances the accuracy and the precision of such algorithm and helps to reduce the unclassified nodes.Second, we propose a hybrid solution based on iterative majority voting and our LiCS. Experiments on real data proves that it outperforms discussed algorithms in terms of accuracy, precision and sensitivity in detecting outliers. It also minimizes the issue of unclassified instances and consolidate the different outputs of algorithms.

Various recommender systems (RSs) have been developed over recent years, and many of them have concentrated on English content. Thus, the majority of RSs from the literature were compared on English content. However, the research investigations about RSs when using contents in other languages such as Arabic are minimal. The researchers still neglect the field of Arabic RSs. Therefore, we aim through this study to fill this research gap by leveraging the benefit of recent advances in the English RSs field. Our main goal is to investigate recent RSs in an Arabic context. For that, we firstly selected five state-of-the-art RSs devoted originally to English content, and then we empirically evaluated their performance on Arabic content. As a result of this work, we first build four publicly available large-scale Arabic datasets for recommendation purposes. Second, various text preprocessing techniques have been provided for preparing the constructed datasets. Third, our investigation derived well-argued conclusions about the usage of modern RSs in the Arabic context. The experimental results proved that these systems ensure high performance when applied to Arabic content.