Skip to Main content Skip to Navigation

Collaviz : un prototype pour la détection et la visualisation de la dynamique collective dans les forums des MOOC

Abstract : Massive Open Online Courses (MOOCs) have seen their numbers increase significantly since the democratization of the Internet. In addition, recently with the COVID-19 pandemic, the trend has intensified. If communication devices such as discussion forums are an integral part of the learning activities of MOOCs, there is still a lack of tools allowing instructors and researchers to guide and finely analyze the learning that takes place there. Dashboards summarizing students' activites are regularly offered to instructors, but they do not allow them to understand collective activities in the forums. From a socio-constructivist point of view, the exchanges and interactions sought by instructors in forums are essential for learning (Stephens, 2014). So far, studies have analyzed interactions in two ways: semantically but on a small scale or statistically and on a large scale but ignoring the quality of the interactions. The scientific contribution of this thesis relates to the proposal of an interactive detection approach of collective activities which takes into account their temporal, semantic and social dimensions. We seek to answer the problem of detecting and observing the collective dynamics that take place in MOOC forums. By collective dynamic, we mean all the qualitative and quantitative interactions of learners in the forums and their temporal changes. We want to allow instructors to intervene to encourage these activities favorable to learning. We rely on studies (Boroujeni 2017, Dascalu 2017) which propose to combine statistical analysis of interactions and automatic language processing to study the flow of information in forums. But, unlike previous studies, our approach is not limited to global or individual-centered analysis. We propose a method of designing indicators and dashboards allowing changes of scales and customization of views in order to support instructors and researchers in their task of detecting, observing and analyzing collective dynamics. To support our approach, we set up questionnaires and conducted semi-structured interviews with the instructors. As for the evaluation of the first indicators built at each iteration of our approach, we used various data sources and formats: Coursera (CSV), Hangout (JSON), Moodle (SQL).
Complete list of metadata
Contributor : Abes Star :  Contact
Submitted on : Thursday, March 4, 2021 - 3:25:10 PM
Last modification on : Friday, March 5, 2021 - 12:19:24 PM
Long-term archiving on: : Saturday, June 5, 2021 - 7:03:02 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03159478, version 1



Malik Koné. Collaviz : un prototype pour la détection et la visualisation de la dynamique collective dans les forums des MOOC. Informatique et langage [cs.CL]. Université du Maine; Institut National Polytechnique Félix Houphouët-Boigny (Yamoussoukro), 2020. Français. ⟨NNT : 2020LEMA1029⟩. ⟨tel-03159478⟩



Record views


Files downloads