Skip to Main content Skip to Navigation
Master Thesis

Développement d'un classifieur hybride pour le domaine des « Ressources Humaines »

Abstract : In this work, we present an application of an automatic text classification system. In order to do so, we developed a product for a company that will be applied to a customer case : an automatic classifier for human resources survey responses, based on a hybrid system (machine learning on annotated data, coupled with expert rules and lexicons). The first approach is based on statistical techniques. It consists on performing multi-categorical annotation to train the model in order to classify the text. The second approach is symbolic. A taxonomy of the humain ressources domain has been developed for the establishment of semantic annotation on texts. We, then, developed a hybrid classification system. At the end, we present the results of evaluation measures to determine the performance of the classifier on our corpus.
Document type :
Master Thesis
Complete list of metadatas

https://dumas.ccsd.cnrs.fr/dumas-03018879
Contributor : Uga - Bulles <>
Submitted on : Monday, November 23, 2020 - 10:17:33 AM
Last modification on : Thursday, November 26, 2020 - 3:17:52 AM

File

GAFSI Myriam_M2_IDL.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-03018879, version 1

Citation

Myriam Gafsi. Développement d'un classifieur hybride pour le domaine des « Ressources Humaines ». Sciences de l'Homme et Société. 2020. ⟨dumas-03018879⟩

Share

Metrics

Record views

25

Files downloads

11