https://doi.org/10.1140/epjds/s13688-026-00648-z
Research
Dual-input deep learning–based approach for propaganda narratives detection in a low-resource language: a case study in Lithuanian
Institute of Data Science and Digital Technologies, Vilnius University, Akademijos g. 4, 08412, Vilnius, Lithuania
a This email address is being protected from spambots. You need JavaScript enabled to view it.
, This email address is being protected from spambots. You need JavaScript enabled to view it.
Received:
13
October
2025
Accepted:
17
March
2026
Published online:
26
March
2026
Abstract
Propaganda narratives play a central role in disseminating disinformation and are tailored to each targeted country’s context and native language. This highlights the pressing need for automated systems capable of detecting and analyzing such narratives, particularly in low-resource languages, where natural language processing tools and data resources remain limited. In this study, we introduce the first supervised machine learning system for identifying pro-Kremlin propaganda narratives in Lithuanian, a low-resource language spoken in a country targeted by Kremlin disinformation. To our knowledge, this represents the first such approach not only for Lithuanian but also for other languages in Russia’s broader neighborhood. Our method employs a novel dual-input architecture that significantly outperforms single-input baselines across all analyzed narratives and even surpasses the performance of ChatGPT-5. Although our study focuses on Lithuanian, our findings and methodology are applicable to other low-resource languages, offering practical guidelines for extending propaganda-narrative detection globally.
Key words: Transformers / Hybrid approach / Low-resource language / Propaganda / Deep-learning / Narratives detection
Handling Editor: Diogo Pacheco
© The Author(s) 2026
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
