eXplaining Competency and Autonomy development in Learning Environments – xCALE
xCALE eXplaining Competency and Autonomy development in Learning Environments
Our main objective in this project is to investigate how to support successfully self-regulated learning at a large scale, with an approach that aims to estimate online acquired skills levels and metacognitive levels about students to provide appropriate interventions.
With the advent of MOOCs, and the success of exercise-based platforms, large-scale online environments are becoming widespread, both in distant learning and in blended learning. Self-Regulated Learning (SRL) is known to have a good potential on autonomy development and on maintaining motivation for learners, both in MOOCs, and in blended learning within exercise-based platforms. Developing Self-Regulated Learning Strategies is also known to have a positive impact on academic achievement. Our main objective in this project is to investigate how to support successfully self-regulated learning at a large scale, with an approach that aims to estimate acquired skills levels and metacognitive levels about students to provide appropriate interventions. <br />This approach pursues the objective of fostering students’ autonomy through Open Learner Models (OLM). Along the same lines as Conati et al. (2018), who demonstrates that effective support assumes that models are interpretable by users, we defend that support should be able to provide some forms of explanation on learning process to stimulate reflection on metacognition, beyond simple activity recommendations. We are interested in Bayesian modeling techniques to acquire and update learners’ models as OLMs (about their cognitive and metacognitive skills and progression), with an application on data corpus resulting from available large-scale learning platforms comprising MOOCs series and an exercise-based platform. We expect from this project a good potential for generalizability and transferability. These MOOCs include exercise databases, different use cases and aim at proposing personalized learning paths based on competencies levels. <br />In concrete terms, xCALE project proposes to develop, experiment and evaluate a generic approach that allows to provide a personalized and interpretable support for (i) skills acquisition and (ii) self regulated learning to support learners’ autonomy. Personalized learning and self-regulated learning will be based on learners’ models - having features for personalization and for self-regulation - and Bayesian modeling techniques.
We are aiming for an agile and iterative approach, to trigger the modelling, development and experimentation processes, oriented towards the project objectives.
- Pedagogical setup will progressively provide refined SRL support and interventions.
- Bayesian Models will first measure cognitive learning, then include progressively SRL strategies and provide related measures.
- Platform development will provide data management and access, and interactions support related to pedagogical setup and bayesian models
- Experimentation on testfields will provide users needs and feedbacks. Testfields will progressively move from small cohorts to get qualitative results to large cohorts to provide quantitative results.
xCALE project proposes to develop, experiment and evaluate a generic approach that allows to provide a personalized and interpretable support for (i) skills acquisition and (ii) self regulated learning to support learners’ autonomy. Personalized learning and self-regulated learning will be based on learners’ models - having features for personalization and for self-regulation - and Bayesian modelling techniques. Expected results and verifiable indicators include what follows:
· A methodology to develop and evaluate Open Learner Models on learners’ skill levels in didactically well-defined disciplines that provide personalized and interpretable insights to users. Proven examples of personalized interpretable interventions in programming and algorithmic courses, based on learner data and expert knowledge practitioners will be provided.
· A generic model on metacognitive processes, based on learners’ progression and SRL-based learning interactions. This model will predict metacognitive strategies to provide relevant interventions.
· A recommendation engine that will provide, on the one hand, personalized activities for skills acquisition according to SRL support, and on the other hand, relevant and personalized interventions to develop SRL (including visual dashboards, dedicated tools, such as time planning or goals management, guidance, etc.).
· An impact study on the transformation of teaching practices, as well as the dissemination of SRL support proposals among other courses and disciplines.
This project addresses issues about education and training, with the aim to contribute to research in large scale learning platforms such as MOOCs and online exercise platforms. This will have implications on advances in the field of Self-Regulated Learning and Artificial Intelligence enhanced learning. We expect from this project many positive effects on the way new generations learn, by providing useful tools enhancing students’ autonomy and engagement as well as adaptive learning.
In short and intermediate terms, this work will have implications on learning and teaching Computer Science in Secondary Education, and in MOOCs, with the learning materials deployed by France-IOI, responding to the lack of teaching resources and tools in Computer Science.
Moreover, an interesting knowledge of the instrumented learning process that this research will provide is a contribution for documenting the training of educators. As recent events show, the global community of educators grappling with the impact of Coronavirus on institutions and student communities, transforming educators and institution practices is at stake. The capacity to design and support a distance learning relationship is increasingly needed. This project is responsive to the actual needs for improving the quality of training hybridization, as it is shown by the social urgency induced by the Coronavirus.
In the long term, this work may impact higher and second education on a broader level. Our research findings and conclusions will be extended and generalized to cover other learning domains and go beyond the Computer Science field. This project has also a good potential and a good foundation for generalizability and transferability, since we dispose of data from large populations, to conduct experimental research. More importantly, comprehensive experiments on a large population of students may reveal the way of SRL, enhanced by our approach and findings, can be fostered in open online environments.
Publications :
Djelil, F., Gilliot, J. M., Garlatti, S., & Leray, P. (2021, August). Supporting Self-Regulation Learning Using a Bayesian Approach. Some Preliminary Insights. In International Joint Conference on Artificial Intelligence IJCAI-21, Workshop Artificial Intelligence for Education.
Roche, M., Gilliot, J.-M., Pentecouteau, H., Lameul, G., Bertrand, E., et Eneau, J. (septembre, 2022). L’autorégulation des apprentissages dans une formation pour adulte. L’exemple de la demande d’aide avec l’utilisation d’une plateforme d’apprentissage [Communication orale]. Colloque AREF, Lausanne, Suisse.
Roche, M., Gilliot, J.-M., Pentecouteau, H., Lameul, G., Bertrand, E., et Eneau, J. (novembre, 2022). L’autorégulation environnementale dans les formations en ligne en France [Communication orale]. Forum Citoyen international de l’Éducation, Tunis, Tunisie.
With the advent of MOOCs, and the success of exercise-based platforms, large-scale online environments are becoming widespread, both in distant learning and in blended learning. Self-Regulated Learning (SRL) is known to have a good potential on autonomy development and on maintaining motivation for learners, both in MOOCs , and in blended learning within exercise-based platforms. Developing Self-Regulated Learning Strategies is also known to have a positive impact on academic achievement. Our main objective in this project is to investigate how to support successfully self-regulated learning at a large scale, with an approach that aims to estimate acquired skills levels and metacognitive levels about students to provide appropriate interventions.
This approach pursues the objective of fostering students’ autonomy through Open Learner Models (OLM). Along the same lines as Conati et al. (2018), who demonstrates that effective support assumes that models are interpretable by users, we defend that support should be able to provide some forms of explanation on learning process to stimulate reflection on metacognition, beyond simple activity recommendations. We are interested in Bayesian modelling techniques to acquire and update learners’ models as OLMs (about their cognitive and metacognitive skills and progression), with an application on data corpus resulting from available large-scale learning platforms comprising MOOCs series (IMT Altantique courses available on FUN and Edx platforms) and courses from a web platform dedicated to learning programming with practical exercises in Secondary Education (France IOI). We expect from this project a good potential for generalizability and transferability, since we dispose of real life data.
In concrete terms, xCALE project proposes to develop, experiment and evaluate a generic approach that allows to provide a personalized and interpretable support for (i) skills acquisition and (ii) self-regulated learning to support learners’ autonomy. Personalized learning and self-regulated learning will be based on learners’ models - having features for personalization and for self-regulation - and Bayesian modelling techniques. Expected results and verifiable indicators include what follows:
- A methodology to develop and evaluate Open Learner Models on learners’ skill levels in didactically well-defined disciplines that provide personalized and interpretable insights to users. Proven examples of personalized interpretable interventions in programming and algorithmic courses, based on learner data and expert knowledge practitioners will be provided.
- A generic model on metacognitive processes, based on learners’ progression and SRL-based learning interactions. This model will predict metacognitive strategies to provide relevant interventions.
- A recommendation engine that will provide, on the one hand, personalized activities for skills acquisition according to SRL support, and on the other hand, relevant and personalized interventions to develop SRL (including visual dashboards, dedicated tools, such as time planning or goals management, guidance, etc.).
- An impact study on the transformation of teaching practices, as well as the dissemination of SRL support proposals among other courses and disciplines.
Project coordination
Jean-Marie Gilliot (Laboratoire des Sciences et Techniques de l'Information, de la Communication et de la Connaissance)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
CREAD Centre de recherche sur l'Education, les Apprentissages et la Didactique
LAB-STICC Laboratoire des Sciences et Techniques de l'Information, de la Communication et de la Connaissance
ASSOCIATION FRANCE-IOI
LS2N Laboratoire des Sciences du Numérique de Nantes
Help of the ANR 483,394 euros
Beginning and duration of the scientific project:
March 2021
- 42 Months
