In collaboration with computational linguist Tsvetomila Mihaylova, and using Bulgarian-language models developed at the GATE Institute in Sofia, the project analysed a corpus of collected headlines and articles. These experiments spanned machine-generated text detection, disinformation likelihood scoring, and sentiment analysis. The film treats their outputs not as evidence, but as unstable readings — surfaces that expose the limits, biases, and ambiguities of automated interpretation.
Within the film, documentary material is translated into speculative cinematic form through collected narratives, mediated traces, and layered processes of transformation. The tension between raw documentary trace and reconstructed mediation mirrors the very conditions the film investigates: an informational environment in which authenticity is perpetually produced, circulated, and destabilized.
CREDITS
A film by Veneta Androva
Machine Learning Support: Tsvetomila Mihaylova
Bulgarian-language-models: The Big Data for Smart Society Institute (GATE)
Sound Design: Daniel Windisch
Voice Contributors: Marta Djourina, Asparuh Donchev, Kiril Anchev, Bojidar Kassabov, Liusi Stavreva
Expert Consultation: Veronika Dimitrova, Martin Stamenov
Subtitle Adaptation: Rosanna Lovell, Alice Escher
Audio Mastering: Sibin Vassilev
TECHNICAL INFORMATION
Duration: 15:58
Country: Bulgaria / Germany
Language: Bulgarian
Subtitles: English
Format: 4K DCP / ProRes
SUPPORTED BY
German Federal Film Board (FFA)
The Bulgarian Fund for Women (BFW)
More Info
0049 30 24045860