Skip to Main content Skip to Navigation
Master Thesis

Prédiction de la difficulté de compréhension des contenus audiovisuels

Abstract : Learning a language is an exercise that requires a lot of effort on the part of the learners. Language teachers are equipped with several supports and tools to help them. One of the most widely used teaching aids is video with excerpts from films. However, this exercise requires teachers to know the level of difficulty of each extract used.
During my internship, I worked on developing a system to automatically annotate the level of difficulty of large amounts of data which would then be used to develop a system that would allow teachers to automatically find movies extracts of the same level as that of their students. For this, I first worked on the definition of syntactic rules from books dedicated to the learning of French as a foreign language and based on the levels of the Common European Framework of Reference for Languages (CEFRL). Then, in order to be able to return labels of levels of difficulty according to defined criteria, I implemented these rules according to labeling with a data programming tool, Snorkel. A set of lexical rules apart from the rest was added with the use of a corpus of French word frequencies by language level (FLELEX).
Document type :
Master Thesis
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-02992702
Contributor : Uga - Bulles <>
Submitted on : Friday, November 6, 2020 - 3:01:50 PM
Last modification on : Sunday, November 8, 2020 - 3:17:45 AM

File

REVOL_Justine_M2_IDL.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-02992702, version 1

Citation

Justine Revol. Prédiction de la difficulté de compréhension des contenus audiovisuels. Sciences de l'Homme et Société. 2020. ⟨dumas-02992702⟩

Share

Metrics

Record views

50

Files downloads

14