Conception, déploiement et évaluation d'une plateforme d'apprentissage adaptatif en ligne

Thesis Abstract

The rapid advancement of digital technologies and the increasing demand for personalized learning experiences have underscored the necessity for adaptive online educational platforms that cater to diverse learner profiles and optimize educational outcomes. Despite the proliferation of e-learning systems, many existing platforms lack dynamic customization capabilities and often fail to adequately address the heterogeneity of learners' needs, leading to suboptimal engagement and learning efficacy. This study aims to design, deploy, and evaluate an innovative online adaptive learning platform that dynamically adjusts instructional content based on individual learner performance, preferences, and engagement levels. The specific objectives are to develop a comprehensive framework for adaptive content delivery, implement the platform integrating machine learning algorithms for real-time personalization, and empirically assess its effectiveness in enhancing learners’ academic performance, engagement, and satisfaction across higher education contexts. Employing a mixed-methods approach, the research adopts a quasi-experimental design supplemented by qualitative insights. The study population encompasses 300 undergraduate students enrolled in foundational courses at a major university, who are randomly selected and divided equally into control and experimental groups. The experimental group interacts with the adaptive platform, while the control group uses a traditional static e-learning system. Data collection instruments include pre- and post-test assessments for academic performance, Likert-scale surveys to measure engagement and satisfaction, and semi-structured interviews to explore user experiences. The reliability of quantitative instruments is established through Cronbach’s alpha coefficients exceeding 0.85, while validity is confirmed via expert review and pilot testing. Data analysis employs statistical techniques such as paired t-tests and ANCOVA to examine differences in learning outcomes, alongside thematic analysis for qualitative interview data, providing a comprehensive understanding of the platform's impact. The anticipated findings suggest that learners engaging with the adaptive platform will demonstrate statistically significant improvements in academic performance (p < 0.01), heightened engagement levels, and elevated satisfaction compared to their counterparts using traditional systems. Furthermore, user feedback is expected to reveal increased motivation, perceived relevance of content, and a desire for continued use, thereby validating the platform’s empirical and experiential effectiveness. The analysis is grounded within the theoretical frameworks of Vygotsky’s Zone of Proximal Development and the Cognitive Load Theory, which inform the adaptive instructional strategies and personalization algorithms utilized in the platform. This research significantly contributes to the body of knowledge by providing a validated model for designing and evaluating adaptive e-learning environments rooted in cognitive and developmental learning theories. It offers practical insights for educational technologists, curriculum designers, and policy makers seeking to implement personalized learning solutions within digital education frameworks. Additionally, the integration of machine learning techniques within a pedagogical context advances the discourse on intelligent tutoring systems and adaptive learning technologies. The study concludes that well-designed adaptive platforms can substantially enhance learning outcomes and user satisfaction, promoting broader adoption of personalized digital education. Recommendations include further refinement of adaptive algorithms to accommodate diverse educational settings, longitudinal investigations to assess sustained impacts, and the integration of additional learner data sources for more nuanced personalization. It is suggested that future research explore scalability and interoperability challenges, as well as the ethical implications of data-driven personalization, to foster more inclusive and effective adaptive learning environments in varied educational contexts.