Sur l'application de méthodes textométriques à la construction de critères de classification en analyse des sentiments

Abstract : About the application of textometric methods for developing classification criteria in Sentiment analysis Over the last ten years, NLP has contributed to applied research on subjectivity, especially in applications such as Opinion mining and Sentiment analysis. However, corpus linguistics and textometry have often addressed the issue of subjectivity in text. Our purpose is to show, first, what textometric analysis could bring to sentiment analysis, and second, the benefits of pooling linguistic/textometric analysis and automatic classification methods based on supervised learning. By processing a corpus of posts from fora, we will show that the building of criteria from a textometric analysis could improve classification results, compared to a purely lexical approach.
Complete list of metadatas

Cited literature [17 references]  Display  Hide  Download

https://halshs.archives-ouvertes.fr/halshs-00702366
Contributor : Mathieu Valette <>
Submitted on : Wednesday, May 30, 2012 - 9:49:07 AM
Last modification on : Monday, July 15, 2019 - 11:10:03 AM
Long-term archiving on : Thursday, December 15, 2016 - 9:35:55 AM

File

TALN2012029.pdf
Publisher files allowed on an open archive

Identifiers

  • HAL Id : halshs-00702366, version 1

Collections

Citation

Egle Eensoo, Mathieu Valette. Sur l'application de méthodes textométriques à la construction de critères de classification en analyse des sentiments. TALN 2012, Jun 2012, Grenoble, France. pp.367-374. ⟨halshs-00702366⟩

Share

Metrics

Record views

476

Files downloads

458