TeAching and Learning InteractionS for Multimodal ANalysis (TALISMAN)
| Funded by : | ANR |
|---|---|
| Funding : | 700 000€ |
| Period : | 2022-2026 |
| Status : | In progress |
| Website : | http://talisman.m-psi.fr/ |
| Coordinator : | Dominique Vaufreydaz |
| Collaborators : | Frédérique Letué, Marie-José Martinez, Jean-Charles Quinton, Francis Jambon, Éric Castelli, Vincent Lestideau (LIG), Romain Laurent, Salomé Cojean, Philippe Dessus, Vasiliki Markaki-Lothe, Laurent Lardy (LaRAC), Christine Michel, Jean-François Cerisier, Hassina El Kechai, Aurélien Nguyen, Emilie Besneville (Techné, Poitiers) |
Description
The aim of the interdisciplinary TALISMAN project is to propose a framework, sharable data and tools to study higher education learning situations including face-to-face, distant and hybrid teaching. Keeping ethics and privacy as main concerns of our research, analyses will be guided by the qualification of the teaching and learning experience, and participants’ engagement. To do so, the project relies on a mixed statistical/machine learning approach to genuine instructional events labeling, including raw events and high-level strategies annotation. The experiments will be undertaken using two existing Context Aware Classrooms (CAC) in Grenoble and Poitiers. Three interrelated research questions about how a CAC, equipped with a variety of unobtrusive multimodal and multilevel sensors, and effectors, is an environment allowing researchers to answer the following questions:
- Q1: How to model and analyze the students’ engagement levels induced by instructional situations, and to formulate prescriptions to teachers and researchers, for teacher development and educational research purposes?
- Q2: How to gather data to build corpora of these different instructional situations and to (semi-) automatically annotate them in a privacy-safe and ethical way?
- Q3: How to devise novel perception algorithms relying on machine learning and statistical techniques and compare their outputs to those delivered by Q1 answers?
Outcomes of the project will include anonymized datasets, source codes and models, description of processes, indicators and protocols used in the project, as well as design and implementation guidelines for higher education courses.