Comparing Multilingual Comparable Articles Based On Opinions

Motaz Saad 1 David Langlois 1 Kamel Smaïli 1
1 PAROLE - Analysis, perception and recognition of speech
Inria Nancy - Grand Est, LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
Abstract : Multilingual sentiment analysis attracts increased attention as the massive growth of multilingual web contents. This conducts to study opinions across different languages by comparing the underlying messages written by different people having different opinions. In this paper, we propose Sentiment based Comparability Measures (SCM) to compare opinions in multilingual comparable articles without translating source/target into the same language. This will allow media trackers (journalists) to automatically detect public opinion split across huge multilingual web contents. To develop SCM, we need either to get or to build parallel sentiment corpora. Because this kind of corpora are not available, we decided to build them. For that, we propose a new method to automatically label parallel corpora with sentiment classes. Then we use the extracted parallel sentiment corpora to develop multilingual sentiment analysis system. Experimental results show that, the proposed measure can capture differences in terms of opinions. The results also show that comparable articles variate in their objectivity and positivity.
Document type :
Conference papers
Liste complète des métadonnées

Cited literature [15 references]  Display  Hide  Download

https://hal.inria.fr/hal-00851959
Contributor : Motaz Saad <>
Submitted on : Monday, August 19, 2013 - 1:12:16 PM
Last modification on : Tuesday, December 18, 2018 - 4:38:02 PM
Document(s) archivé(s) le : Wednesday, November 20, 2013 - 4:14:59 AM

File

Saad_BUCC_ACL_2013.pdf
Publisher files allowed on an open archive

Identifiers

  • HAL Id : hal-00851959, version 1

Collections

Citation

Motaz Saad, David Langlois, Kamel Smaïli. Comparing Multilingual Comparable Articles Based On Opinions. Proceedings of the 6th Workshop on Building and Using Comparable Corpora, Association for Computational Linguistics ACL, Aug 2013, Sofia, Bulgaria. pp.105-111. ⟨hal-00851959⟩

Share

Metrics

Record views

620

Files downloads

507