Machine unlearning · mechanistic interpretability · information integrity
Research
How do data systems remember, forget, and represent the world?
I study how mechanistic interpretability can inform unlearning, alongside work on narrative unlearning, bias mitigation, Ukrainian NLP, and participatory data modeling.
Listed by publication month, newest first. Presentation dates are given separately; the master’s thesis record specifies only the year.
2025–2026
Publications & preprints
Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS
Presented at WIPE-OUT 2 — 2nd Workshop on Machine Unlearning and Privacy Preservation, ECML PKDD 2026 · 7 September 2026 · Naples, Italy.
Preprint: arXiv:2607.22657 · revised 31 July 2026.
LENS tests whether models reproduce targeted narratives across different prompting contexts, measuring suppression alongside degraded outputs.
Citation
Makovska, V., & Fletcher, G. (2026). Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS. arXiv:2607.22657. https://arxiv.org/abs/2607.22657
BibTeX
@misc{makovska2026lens,
title = {Between Suppression and Collapse: Evaluating Narrative Unlearning with LENS},
author = {Makovska, Viktoriia and Fletcher, George},
year = {2026},
eprint = {2607.22657},
archivePrefix = {arXiv},
primaryClass = {cs.CL},
url = {https://arxiv.org/abs/2607.22657}
}Graph-Based Detection of Disinformation Narrative Diffusion between Russian and Ukrainian Telegram Channels
In Proceedings of the Fifth Ukrainian Natural Language Processing Conference (UNLP 2026), pages 80-96, Lviv, Ukraine. Association for Computational Linguistics.
This work combines weak supervision with propagation-graph analysis to group semantically related claims and study how disinformation narratives move across Russian-language and Ukrainian-language Telegram channels.
Citation
Vistak, Y., Makovska, V., Schmitt, V., & Solopova, V. (2026). Graph-Based Detection of Disinformation Narrative Diffusion between Russian and Ukrainian Telegram Channels. In Proceedings of the Fifth Ukrainian Natural Language Processing Conference (UNLP 2026), 80-96. Association for Computational Linguistics. https://aclanthology.org/2026.unlp-1.9/
BibTeX
@inproceedings{vistak-etal-2026-graph,
title = {Graph-Based Detection of Disinformation Narrative Diffusion between {R}ussian and {U}krainian {T}elegram Channels},
author = {Vistak, Yuliia and Makovska, Viktoriia and Schmitt, Vera and Solopova, Veronika},
booktitle = {Proceedings of the Fifth {U}krainian Natural Language Processing Conference ({UNLP} 2026)},
year = {2026},
address = {Lviv, Ukraine},
publisher = {Association for Computational Linguistics},
pages = {80--96},
url = {https://aclanthology.org/2026.unlp-1.9/}
}
Data-Efficient Adaptation of Multilingual LLMs to Ukrainian
In Proceedings of the Fifth Ukrainian Natural Language Processing Conference (UNLP 2026), pages 155-168, Lviv, Ukraine. Association for Computational Linguistics.
A reproducible data-centric pipeline for adapting multilingual language models to Ukrainian through vocabulary surgery, cross-lingual quality filtering, and translated and synthetic instruction data.
Citation
Paniv, Y., Didenko, B., Haltiuk, M., Humennyy, V., Kravchenko, A., Kyslyi, R., Makovska, V., Orlovskyi, A., Ruban, B., Rudko, M.-Y., Senyk, A., Drushchak, N., Chaplynskyi, D., & Romanyshyn, M. (2026). Data-Efficient Adaptation of Multilingual LLMs to Ukrainian. In Proceedings of the Fifth Ukrainian Natural Language Processing Conference (UNLP 2026), 155-168. Association for Computational Linguistics. https://aclanthology.org/2026.unlp-1.14/
BibTeX
@inproceedings{paniv-etal-2026-data,
title = {Data-Efficient Adaptation of Multilingual {LLM}s to {U}krainian},
author = {Paniv, Yurii and Didenko, Bohdan and Haltiuk, Mykola and Humennyy, Vladyslav and Kravchenko, Andrian and Kyslyi, Roman and Makovska, Viktoriia and Orlovskyi, Artem and Ruban, Bohdan and Rudko, Maksym-Yurii and Senyk, Anastasiia and Drushchak, Nazarii and Chaplynskyi, Dmytro and Romanyshyn, Mariana},
booktitle = {Proceedings of the Fifth {U}krainian Natural Language Processing Conference ({UNLP} 2026)},
year = {2026},
address = {Lviv, Ukraine},
publisher = {Association for Computational Linguistics},
pages = {155--168},
url = {https://aclanthology.org/2026.unlp-1.14/}
}
Seasoning Data Modeling Education with GARLIC: A Participatory Co-Design Framework
In Proceedings of the Workshops of the EDBT/ICDT 2026 Joint Conference, DataEd'26: 5th International Workshop on Data Systems Education, Tampere, Finland. CEUR Workshop Proceedings, Vol. 4192. Published April 2026; workshop held 24 March 2026.
GARLIC adapts the ONION framework into a workshop-based method for teaching participatory entity-relationship modeling through role-play, collaborative synthesis, critique, and iteration.
Citation
Makovska, V., Michurin, I., Tokhtamysh, M., Fletcher, G., & Stoyanovich, J. (2026). Seasoning Data Modeling Education with GARLIC: A Participatory Co-Design Framework. In Proceedings of the Workshops of the EDBT/ICDT 2026 Joint Conference, DataEd'26: 5th International Workshop on Data Systems Education. CEUR Workshop Proceedings, Vol. 4192. https://ceur-ws.org/Vol-4192/DataEd-paper4.pdf
BibTeX
@inproceedings{Makovska2026GARLIC,
title = {Seasoning Data Modeling Education with GARLIC: A Participatory Co-Design Framework},
author = {Makovska, Viktoriia and Michurin, Ihor and Tokhtamysh, Mariia and Fletcher, George and Stoyanovich, Julia},
booktitle = {Proceedings of the Workshops of the EDBT/ICDT 2026 Joint Conference},
series = {CEUR Workshop Proceedings},
volume = {4192},
year = {2026},
address = {Tampere, Finland},
url = {https://ceur-ws.org/Vol-4192/DataEd-paper4.pdf},
eprint = {2602.18274},
archivePrefix = {arXiv},
primaryClass = {cs.DB}
}
Bridging Applied Experience and Research Contexts in Ukrainian NLP Education
In Proceedings of the Seventh Workshop on Teaching Natural Language Processing (TeachNLP 2026), pages 78-83, Rabat, Morocco. Association for Computational Linguistics.
A report on building an open, bachelor-level NLP course taught in Ukrainian, with culturally adapted tasks, integrated ethics, project-based assessment, and publicly available materials.
Citation
Paniv, Y., & Makovska, V. (2026). Bridging Applied Experience and Research Contexts in Ukrainian NLP Education. In Proceedings of the Seventh Workshop on Teaching Natural Language Processing (TeachNLP 2026), 78-83. Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.teachingnlp-1.12
BibTeX
@inproceedings{paniv-makovska-2026-bridging-applied,
title = {Bridging Applied Experience and Research Contexts in {U}krainian {NLP} Education},
author = {Paniv, Yurii and Makovska, Viktoriia},
booktitle = {Proceedings of the Seventh Workshop on Teaching Natural Language Processing ({T}each{NLP} 2026)},
year = {2026},
address = {Rabat, Morocco},
publisher = {Association for Computational Linguistics},
pages = {78--83},
doi = {10.18653/v1/2026.teachingnlp-1.12},
url = {https://aclanthology.org/2026.teachingnlp-1.12/}
}
Memory Undone: Between Knowing and Not Knowing in Data Systems
arXiv preprint arXiv:2602.21180 [cs.CY], submitted February 24, 2026. Presented at Undone Computer Science on 24 March 2026.
This position paper distinguishes erasure, unlearning, exclusion, and forgetting, and argues that machine unlearning should be evaluated not only by model utility but also by transparency, accountability, and governance.
Citation
Makovska, V., Fletcher, G., Stoyanovich, J., & Zakharchenko, T. (2026). Memory Undone: Between Knowing and Not Knowing in Data Systems. arXiv preprint arXiv:2602.21180. https://arxiv.org/abs/2602.21180
BibTeX
@misc{Makovska2026MemoryUndone,
title = {Memory Undone: Between Knowing and Not Knowing in Data Systems},
author = {Makovska, Viktoriia and Fletcher, George and Stoyanovich, Julia and Zakharchenko, Tetiana},
year = {2026},
eprint = {2602.21180},
archivePrefix = {arXiv},
primaryClass = {cs.CY},
url = {https://arxiv.org/abs/2602.21180}
}
ONION: A Multi-Layered Framework for Participatory ER Design
In Proceedings of the Workshop on Human-In-the-Loop Data Analytics (HILDA '25), June 22, 2025, Berlin, Germany.
ONION is a five-stage participatory method for moving from unstructured stakeholder input to entity-relationship models while making assumptions, omissions, and designer choices easier to inspect.
Citation
Makovska, V., Fletcher, G., & Stoyanovich, J. (2025). ONION: A Multi-Layered Framework for Participatory ER Design. In Proceedings of the Workshop on Human-In-the-Loop Data Analytics (HILDA '25). https://doi.org/10.1145/3736733.3736736
BibTeX
@inproceedings{Makovska2025ONION,
author = {Makovska, Viktoriia and Fletcher, George and Stoyanovich, Julia},
title = {ONION: A Multi-Layered Framework for Participatory ER Design},
booktitle = {Proceedings of the Workshop on Human-In-the-Loop Data Analytics (HILDA '25)},
year = {2025},
doi = {10.1145/3736733.3736736}
}
Vandalism or Propaganda? Enhancing Vandalism Detection in Ukrainian and Russian Wikipedia through Knowledge Manipulation Filtering
6th Masters Symposium on Advances in Data Mining, Machine Learning, and Computer Vision (MS-AMLV 2025), 28–29 March 2025, Lviv, Ukraine.
This work explores how changes in a Russian government-backed Wikipedia fork can provide signals for identifying knowledge manipulation in Ukrainian and Russian Wikipedia edits.
Citation
Makovska, V., Trokhymovych, M., & Saez-Trumper, D. (2025). Vandalism or Propaganda? Enhancing Vandalism Detection in Ukrainian and Russian Wikipedia through Knowledge Manipulation Filtering. 6th Masters Symposium on Advances in Data Mining, Machine Learning, and Computer Vision (MS-AMLV 2025). https://apps.ucu.edu.ua/wp-content/uploads/2025/07/MS-AMLV-2025-CR-Submission-No-9.pdf
BibTeX
@misc{makovska2025vandalismpropaganda,
title = {Vandalism or Propaganda? Enhancing Vandalism Detection in Ukrainian and Russian Wikipedia through Knowledge Manipulation Filtering},
author = {Makovska, Viktoriia and Trokhymovych, Mykola and Saez-Trumper, Diego},
year = {2025},
note = {MS-AMLV 2025 symposium paper, Ukrainian Catholic University, Lviv},
url = {https://apps.ucu.edu.ua/wp-content/uploads/2025/07/MS-AMLV-2025-CR-Submission-No-9.pdf}
}Open work
Projects & resources
Unlearning for disinformation and bias mitigation
As a Research Engineer at Lapa LLM, I work on disinformation mitigation strategies, narrative unlearning, and bias mitigation through unlearning.
Participatory data modeling workshop materials
Open scenario cards, role cards, and facilitation instructions for teaching participatory entity-relationship modeling.
Ukrainian bias-evaluation datasets
I co-developed the Ukrainian translation and adaptation of StereoSet with Oleksandra Ostafiichuk and Tetiana Zakharchenko.
FairForget
With Tetiana Zakharchenko, I investigate fairness and hidden biases after machine unlearning. Our project received support through Women Led Science.
Foundations
Thesis & PhD proposal
Forgetting Machines: Responsibility by Design in Machine Unlearning
PhD Topic Proposal, UCU Applied Sciences Faculty (2025).
Citation
Makovska, V. (2025). Forgetting Machines: Responsibility by Design in Machine Unlearning [PhD topic proposal]. Ukrainian Catholic University.
Vandalism or Knowledge Manipulation? Detecting Narratives in Wikipedia Edits
Master's thesis, Ukrainian Catholic University, Faculty of Applied Sciences (2025).
Citation
Makovska, V. (2025). Vandalism or Knowledge Manipulation? Detecting Narratives in Wikipedia Edits (Master's thesis). Ukrainian Catholic University.