Proceedings of the International Conference on Social Science, Psychology and Legal Regulation (SPL 2021)

Sentiment Analysis Based on the BERT Model: Attitudes Towards Politicians Using Media Data

Authors
Svitlana Salnikova*, ORCID, Roman KyrychenkoORCID
Taras Shevchenko National University of Kyiv, Kyiv, Ukraine
Corresponding Author
Svitlana Salnikova
Available Online 18 December 2021.
DOI
10.2991/assehr.k.211218.007How to use a DOI?
Keywords
computational text analysis; targeted-BERT; tone dynamics; neural network; model training
Abstract

The latest analysis methods of sentiments borrowed from computational linguistics are relevant in the age of big data, which is difficult to process through traditional content analysis. These methods have made it possible to analyze information over a long period, which allows us to trace the dynamics of the relationship to a particular object over time and large-scale comparative studies of texts. The authors demonstrate the applicability of sentiment analysis based on transformer models to the study of the temporal model of attitudes towards well-known politicians (2001-2021) on the example of text analysis of multilingual online publications. To do this, the authors used the targeted-BERT method for automated directed analysis of sentiments, obtained quality indicators F1-score 0.799 and 0.741 for Ukrainian and Russian models, respectively. The authors tested the dependence of mediatization of politicians on the country’s political hierarchy, confirmed hypotheses about the attitude to their power (more significant criticism of the Ukrainian media and gradual loyalty to the Russian media) and foreign politicians (dominance of negative tone in both media with a growing trend for Ukrainian media).

Copyright
© 2021 The Authors. Published by Atlantis Press SARL.
Open Access
This is an open access article under the CC BY-NC license.

Download article (PDF)

Volume Title
Proceedings of the International Conference on Social Science, Psychology and Legal Regulation (SPL 2021)
Series
Advances in Social Science, Education and Humanities Research
Publication Date
18 December 2021
ISBN
978946239-489-6
ISSN
2352-5398
DOI
10.2991/assehr.k.211218.007How to use a DOI?
Copyright
© 2021 The Authors. Published by Atlantis Press SARL.
Open Access
This is an open access article under the CC BY-NC license.

Cite this article

TY  - CONF
AU  - Svitlana Salnikova
AU  - Roman Kyrychenko
PY  - 2021
DA  - 2021/12/18
TI  - Sentiment Analysis Based on the BERT Model: Attitudes Towards Politicians Using Media Data
BT  - Proceedings of the International Conference on Social Science, Psychology and Legal Regulation (SPL 2021)
PB  - Atlantis Press
SP  - 39
EP  - 44
SN  - 2352-5398
UR  - https://doi.org/10.2991/assehr.k.211218.007
DO  - 10.2991/assehr.k.211218.007
ID  - Salnikova2021
ER  -