Skip to Main content Skip to Navigation
Theses

Détection des influenceurs dans des médias sociaux

Abstract : In this thesis, we present the design and evaluation of a system to automatically detect influencers in social media, based on the manifestations of their influencing action in interpersonal communications. Approaches to influencer detection generally use either the structure of communication between individuals, or the analysis of its content. The theoretical framework chosen in our thesis has the particularity of combining these two types of approach for their complementarity. We characterise the action of influencers at the level of a target individual, from its means to its effects, by discursive features of both the messages sent by the influencers and those sent by the influenced individuals. The automatic detection of these discursive features in social media messages is done with methods in natural language processing, based on linguistic rules and machine learning models. At the group level, the action of influencers is characterised by their central position in a social graph, that represents interpersonal actions within the group. The hybridity of our system consists in the use of linguistic information, automatically extracted in discussions, to construct the social graphs whose structure will be analysed.
Document type :
Theses
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-03640442
Contributor : ABES STAR :  Contact
Submitted on : Wednesday, April 13, 2022 - 3:16:23 PM
Last modification on : Tuesday, April 19, 2022 - 10:28:14 AM
Long-term archiving on: : Thursday, July 14, 2022 - 6:45:27 PM

File

DETURCK_kevin_vavd.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03640442, version 1

Collections

Citation

Kévin Deturck. Détection des influenceurs dans des médias sociaux. Ordinateur et société [cs.CY]. Institut National des Langues et Civilisations Orientales- INALCO PARIS - LANGUES O', 2021. Français. ⟨NNT : 2021INAL0034⟩. ⟨tel-03640442⟩

Share

Metrics

Record views

185

Files downloads

162