Disinformation in Scientific News in Bulgarian

Last modified by ruslana m on 2026/09/17 09:01

Published Sunday 11 October 2026 at 09:31

The article by Ruslana Margova examines how pseudoscientific and misleading narratives are embedded in Bulgarian-language science communication. The study analyzes 17,833 posts from Bulgarian public Facebook pages and groups published between 2014 and 2024, collected via CrowdTangle, alongside 498 science-related news articles from early 2026. The methodology combines TF-IDF and n-gram analysis, a specialized lexicon of pseudoscientific markers, expert validation, and thematic clustering using ChatGPT-4o.

A key finding is that 17.68% of the analyzed social media posts contain lexical markers associated with pseudoscientific or disinformation discourse. The most prominent thematic cluster, labelled “Bulgaria Above All – The Most Ancient,” appears in 55.27% of the identified pseudoscientific content. Other major themes include “energy” and “quantum” narratives (40.32%), health and alternative medicine (34.49%), and conspiracy theories and “hidden truths” (32.42%). These categories overlap and therefore the percentages should not be added together.

One of the article’s central observations is the strong connection between pseudoscience and national identity and narratives of historical exceptionalism. Examples include claims about the extraordinary antiquity or uniqueness of Bulgarians and sensationalized accounts of supposedly Bulgarian scientific discoveries. The study also identifies a pattern of pseudo-expertise, in which scientific-sounding terms such as quantum, energy, and frequency are transferred into astrological, esoteric, or otherwise non-scientific contexts.

The comparison between different media environments suggests that social media is more strongly associated with emotional, national-identity, and conspiratorial narratives, whereas the news corpus contains more horoscopes, alternative medicine, and sensationalized representations of scientific topics.

An important methodological contribution is the comparison with ChatGPT-4o. While the model successfully identifies broad themes such as health misinformation, conspiracy theories, and esoteric uses of scientific terminology, it does not identify the specifically Bulgarian nationalist narrative as a dominant category. The author interprets this as a potential “cultural blind spot” in general-purpose LLMs, highlighting the value of combining AI-assisted analysis with locally developed linguistic resources and expert validation.

Suggested citation

Margova, Ruslana. 2026. “Detecting Pseudoscientific Narratives in Bulgarian Science Communication: Nationalistic Mystification, Pseudo-Expertise and Cultural Blind Spots in AI-Assisted Analysis.” Proceedings of the Seventh International Conference on Computational Linguistics in Bulgaria (CLIB 2026), 228–237. Sofia, Bulgaria: Department of Computational Linguistics, Institute for Bulgarian Language, Bulgarian Academy of Sciences.

Article: Full-text PDF

Proceedings: CLIB Conference Proceedings

Co -Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or DIGITAL-2021-TRUST-01. Neither the European Union nor the granting authority can be held responsible for them.

Disinformation is commonly understood as a problem of factual accuracy — the creation and circulation of false or misleading information. This research explores a less visible dimension of the phenomenon: what happens when manipulation begins to affect the shared meanings through which political reality itself is interpreted.

A new study The Semantic Shift of the Concept of Peace during the Russian Invasion of Ukraine, made by BROD researcher and published in the Journal of Contemporary European Studies, approaches disinformation from this perspective by examining changes in the political and media use of one of the most fundamental concepts in international relations: peace.

[+]

The article by Stanislav Penkov, “A WordNet-Aligned Taxonomy of Logical Fallacies for Low-Resource NLP,” proposes a compact and multilingual framework for identifying logical fallacies in natural language, with a particular focus on low-resource languages such as Bulgarian. The main motivation is that existing fallacy-detection datasets use inconsistent classification schemes and are predominantly English-language, making cross-dataset comparison and cross-lingual transfer difficult. [+]

Disinformation in Scientific News in Bulgarian

Suggested citation

BROD

EDMO Network

Follow us