Capitaliser les processus d'analyse de traces d'apprentissage : modélisation ontologique & assistance à la réutilisation

Abstract : This thesis in computer science focuses on the problem of capitalizing analysis processes of elearning traces within the Learning Analytics (LA) community. The aim is to allow these analysis processes to be shared, adapted and reused. Currently, this capitalization is limited by two important factors: the analysis processes are dependent on the analysis tools that implement them - their technical context - and the pedagogical context for which they are conducted. This prevents them from being shared, but also from being simply reused outside their original contexts, even if the new contexts are similar. The objective of this thesis is to provide models and methods for the capitalisation of analysis processes of elearning traces, as well as to assist the various actors involved in the analysis, particularly during the reuse phase. To do this, we answer the following three scientific questions: how to share and combine analysis processes implemented in different analysis tools; how to reuse an existing analysis process to meet another analysis need; how to assist the different actors in the development and exploitation of analysis processes; and how to support them in the development and exploitation of analysis processes. Our first contribution, resulting from a synthesis of the state of the art, is the formalization of a cycle of elaboration and exploitation of the analysis processes, in order to define the different stages, the different actors and their different roles. This formalization is accompanied by a definition of capitalization and its properties. Our second contribution responds to the first barrier related to the technical dependence of current analysis processes and their sharing. We propose a meta-model that allows to describe the analysis processes independently of the analysis tools. This meta-model formalizes the description of the operations used in the analysis processes, the processes themselves and the traces used, in order to avoid the technical constraints caused by these tools. This formalism, common to the analysis processes, also makes it possible to consider their sharing. It has been implemented and evaluated in one of our prototypes. Our third contribution deals with the second lock on the reuse of analysis processes. We propose an ontological framework for analysis processes, which allows semantic elements to be directly introduced, in a structured way, during the description of analysis processes. This narrative approach thus enriches the previous formalism and makes it possible to satisfy the properties of understanding, adaptation and reuse necessary for capitalisation. This ontological approach was implemented and evaluated in another of our prototypes. Finally, our last contribution responds to the last lock identified and concerns new assistances to actors, in particular a new method of researching analysis processes, based on our previous proposals. We use the ontological framework of the narrative approach to define inference rules and heuristics to reason about the analysis processes as a whole (e.g. steps, configurations) during the research. We also use the semantic network underlying this ontological modeling to strengthen assistance to actors by providing them with inspection and understanding tools during the research. This assistance was implemented in one of our prototypes, and empirically evaluated.
Complete list of metadatas

Cited literature [377 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02164400
Contributor : Alexis Lebis <>
Submitted on : Tuesday, June 25, 2019 - 9:42:50 AM
Last modification on : Tuesday, September 17, 2019 - 10:36:41 AM

File

main.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-02164400, version 1

Citation

Alexis Lebis. Capitaliser les processus d'analyse de traces d'apprentissage : modélisation ontologique & assistance à la réutilisation. Environnements Informatiques pour l'Apprentissage Humain. Sorbonne Université, CNRS, Laboratoire d'Informatique de Paris 6, LIP6, Paris, France, 2019. Français. ⟨tel-02164400⟩

Share

Metrics

Record views

111

Files downloads

172