---
title: "Hidetsune at SemEval-2026 Task 10 — paper by Takahashi (2026)"
description: "This system paper addresses SemEval-2026 Task 10 Subtask 2, a binary task of deciding whether a… 17 references, cited by 1 paper here."
authors: ["Hidetsune Takahashi"]
year: 2026
venue: "Proceedings of the 20th International Workshop on Semantic Evaluation (2026)"
source: "https://aclanthology.org/2026.semeval-1.26/"
doi: "10.18653/v1/2026.semeval-1.26"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/takahashi-2026-hidetsune-semeval-2026-task-10/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/takahashi-2026-hidetsune-semeval-2026-task-10/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/13/) / Hidetsune at SemEval-2026 Task 10: A Systematic Exploration of Training and Inference Strategies for Detecting Conspiracy Beliefs

# Hidetsune at SemEval-2026 Task 10: A Systematic Exploration of Training and Inference Strategies for Detecting Conspiracy Beliefs

**Authors:** [Hidetsune Takahashi](https://md.research.elamur.ai/free/en/authors/hidetsune-takahashi/) · **Year:** 2026 · **Venue:** [Proceedings of the 20th International Workshop on Semantic Evaluation (2026)](https://md.research.elamur.ai/free/en/venues/semeval/) ([SemEval 2026](https://md.research.elamur.ai/free/en/venues/semeval/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.semeval-1.26/) · [PDF](https://aclanthology.org/2026.semeval-1.26.pdf) · [DOI 10.18653/v1/2026.semeval-1.26](https://doi.org/10.18653/v1/2026.semeval-1.26) · [OpenAlex](https://openalex.org/W7166832552)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [1](https://md.research.elamur.ai/free/en/papers/takahashi-2026-hidetsune-semeval-2026-task-10/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
This system paper addresses SemEval-2026 Task 10 Subtask 2, a binary task of deciding whether a Reddit comment expresses a conspiracy belief. The author first fine-tuned several open-source models on the hydrated JSONL training data, balancing Yes and No through undersampling. The models were BERT, RoBERTa and TwHIN-BERT (base and large) and two GPT-2 variants. RoBERTa-large was selected as the base model. On top of it, four additions were tested: an instruction prompt prepended to inputs, back-translation augmentation through Spanish, class-weighted cross-entropy, and focal loss. None of these beat plain fine-tuning with undersampling. The final step tuned the decision threshold on the softmax probabilities. With a threshold of 0.4, the system reached about 0.818 weighted F1 on the development set, ranking near the top. On the test set, however, scores dropped by roughly 0.1 and became less sensitive to the threshold. This suggests the two splits differ in distribution. The paper is useful mainly as a negative result: common imbalance and augmentation tricks gave no benefit when class skew is moderate.

## Key points
- The training split has 1,541 Yes, 1,990 No and 785 Can't tell comments. Weighted F1 is the main metric, alongside accuracy and per-class F1 for Yes and No.
- Large encoders worked best, with BERT-large and RoBERTa-large at about 0.79. GPT-2 models scored about 0.60–0.66, and TwHIN-BERT about 0.71–0.73.
- Every added technique lowered weighted F1 compared with the RoBERTa-large baseline:
  - back-translation: 0.771
  - instruction tuning: 0.757
  - weighted CE: 0.754
  - focal loss (γ = 2.0): 0.714
- Threshold tuning on dev produced a peak of 0.818 at 0.4, placing 4th of 45 teams. On test, the best result was 0.734 at a threshold of 0.6, placing 31st of 51 teams.
- Code is released at github.com/Hidetsune/SemEval2026_Task10. The author lists two limitations: little data preprocessing, and no use of the Subtask 1 marker spans.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Waseda University](https://md.research.elamur.ai/free/en/universities/waseda-university/) — University, Tokyo, Japan · 20 works · 584 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 577 works

## Cite

Cite as: Hidetsune Takahashi. 2026. Hidetsune at SemEval-2026 Task 10: A Systematic Exploration of Training and Inference Strategies for Detecting Conspiracy Beliefs. Proceedings of the 20th International Workshop on Semantic Evaluation (2026).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{takahashi2026hidetsune,
  title = {{Hidetsune at SemEval-2026 Task 10: A Systematic Exploration of Training and Inference Strategies for Detecting Conspiracy Beliefs}},
  author = {Hidetsune Takahashi},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.26},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.26}
}
```

- [← previous Hi-SEMFLOW: Lie Algebra–Based Semantic Flow for Span-Level Informal Language Identification in Hindi](https://md.research.elamur.ai/free/en/papers/ravikiran-2026-hi-semflow-lie-algebra-based/)
- [next → Hidetsune at SemEval-2026 Task 11: Adapting Pretrained Reasoning Models with Deep Supervision and Inference Refinement for Content-Independent Validity Classification](https://md.research.elamur.ai/free/en/papers/takahashi-2026-hidetsune-semeval-2026-task-11/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/13/)

## References

17 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Matteo Cinelli, Gabriele Etta, Michele Avalle, Alessandro Quattrociocchi, Niccolò Di Marco, Carlo Valensise, Alessandro Galeazzi, and Walter Quattrociocchi. 2022. Conspiracy theories and social media platforms. Current Opinion in Psychology, 47:101407. [find](https://scholar.google.com/scholar?q=Matteo+Cinelli%2C+Gabriele+Etta%2C+Michele+Avalle%2C+Alessandro+Quattrociocchi%2C+Niccol%C3%B2+Di+Marco%2C+Carlo+Valensise%2C+Alessandro+Galeazzi%2C+and+Walter+Quattrociocchi%2E+2022%2E+Conspiracy+theories+and+social+media+)
2. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. [BERT: Pre-training of deep bidirectional transformers for language understanding](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/). In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,…
3. Karen M Douglas. 2021. Are conspiracy theories harmless? The Spanish journal of psychology, 24:e13. [find](https://scholar.google.com/scholar?q=Karen+M+Douglas%2E+2021%2E+Are+conspiracy+theories+harmless%3F+The+Spanish+journal+of+psychology%2C+24%3Ae13%2E)
4. Karen M Douglas, Robbie M Sutton, and Aleksandra Cichocka. 2017. The psychology of conspiracy theories. Current directions in psychological science, 26(6):538–542. [find](https://scholar.google.com/scholar?q=Karen+M+Douglas%2C+Robbie+M+Sutton%2C+and+Aleksandra+Cichocka%2E+2017%2E+The+psychology+of+conspiracy+theories%2E+Current+directions+in+psychological+science%2C+26%286%29%3A538%E2%80%93542%2E)
5. Karen M Douglas, Joseph E Uscinski, Robbie M Sutton, Aleksandra Cichocka, Turkay Nefes, Chee Siang Ang, and Farzin Deravi. 2019. Understanding conspiracy theories. Political psychology, 40:3–35. [find](https://scholar.google.com/scholar?q=Karen+M+Douglas%2C+Joseph+E+Uscinski%2C+Robbie+M+Sutton%2C+Aleksandra+Cichocka%2C+Turkay+Nefes%2C+Chee+Siang+Ang%2C+and+Farzin+Deravi%2E+2019%2E+Understanding+conspiracy+theories%2E+Political+psychology%2C+40%3A3%E2%80%9335%2E)
6. Sergey Edunov, Myle Ott, Michael Auli, and David Grangier. 2018. Understanding back-translation at scale. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 489–500, Brussels, Belgium. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Sergey+Edunov%2C+Myle+Ott%2C+Michael+Auli%2C+and+David+Grangier%2E+2018%2E+Understanding+back%2Dtranslation+at+scale%2E+In+Proceedings+of+the+2018+Conference+on+Empirical+Methods+in+Natural+Language+Processing%2C+pag)
7. Adam M Enders, Joseph E Uscinski, Michelle I Seelig, Casey A Klofstad, Stefan Wuchty, John R Funchion, Manohar N Murthi, Kamal Premaratne, and Justin Stoler. 2023. The relationship between social media use and beliefs in conspiracy theories and misinformation. Political behavior, 45(2):781–804. [find](https://scholar.google.com/scholar?q=Adam+M+Enders%2C+Joseph+E+Uscinski%2C+Michelle+I+Seelig%2C+Casey+A+Klofstad%2C+Stefan+Wuchty%2C+John+R+Funchion%2C+Manohar+N+Murthi%2C+Kamal+Premaratne%2C+and+Justin+Stoler%2E+2023%2E+The+relationship+between+social+medi)
8. Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. [find](https://scholar.google.com/scholar?q=Alec+Radford%2C+Jeff+Wu%2C+Rewon+Child%2C+David+Luan%2C+Dario+Amodei%2C+and+Ilya+Sutskever%2E+2019%2E+Language+models+are+unsupervised+multitask+learners%2E)
9. Mattia Samory, Felix Soldner, and Veronika Batzdorfer. 2026. [SemEval-2026 task 10: PsyCoMark – psycholinguistic conspiracy marker extraction and detection](https://md.research.elamur.ai/free/en/papers/samory-2026-semeval-2026-task-10-psycomark/). In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
10. Hidetsune Takahashi, Xingru Lu, Sean Ishijima, De-okgyu Seo, Yongju Kim, Sehoon Park, Min Song, Kathylene Marante, Keitaro-luke Iso, Hirotaka Tokura, and Emily Ohman. 2024. OZemi at SemEval-2024 task 1: A simplistic approach to textual relatedness evaluation using transformers and machine translatio… [find](https://scholar.google.com/scholar?q=Hidetsune+Takahashi%2C+Xingru+Lu%2C+Sean+Ishijima%2C+De%2Dokgyu+Seo%2C+Yongju+Kim%2C+Sehoon+Park%2C+Min+Song%2C+Kathylene+Marante%2C+Keitaro%2Dluke+Iso%2C+Hirotaka+Tokura%2C+and+Emily+Ohman%2E+2024%2E+OZemi+at+SemEval%2D2024+task+)
11. Hidetsune Takahashi, Sumiko Teng, Jina Lee, Wenxiao Hu, Rio Obe, Chuen Shin Yong, and Emily Ohman. 2025. OZemi at SemEval-2025 task 11: Multilingual emotion detection and intensity. In Proceedings of the 19th International Workshop on Semantic Evaluation (SemEval-2025), pages 109–115, Vienna, Austri… [find](https://scholar.google.com/scholar?q=Hidetsune+Takahashi%2C+Sumiko+Teng%2C+Jina+Lee%2C+Wenxiao+Hu%2C+Rio+Obe%2C+Chuen+Shin+Yong%2C+and+Emily+Ohman%2E+2025%2E+OZemi+at+SemEval%2D2025+task+11%3A+Multilingual+emotion+detection+and+intensity%2E+In+Proceedings+of+)
12. Jörg Tiedemann and Santhosh Thottingal. 2020. OPUS-MT – building open translation services for the world. In Proceedings of the 22nd Annual Conference of the European Association for Machine Translation, pages 479–480, Lisboa, Portugal. European Association for Machine Translation. [find](https://scholar.google.com/scholar?q=J%C3%B6rg+Tiedemann+and+Santhosh+Thottingal%2E+2020%2E+OPUS%2DMT+%E2%80%93+building+open+translation+services+for+the+world%2E+In+Proceedings+of+the+22nd+Annual+Conference+of+the+European+Association+for+Machine+Translati)
13. Xinyang Zhang, Yury Malkov, Omar Florez, Serim Park, Brian McWilliams, Jiawei Han, and Ahmed El-Kishky. 2022. Twhin-bert: A socially-enriched pre-trained language model for multilingual tweet representations. arXiv preprint arXiv:2209.07562. — [arXiv:2209.07562](https://arxiv.org/abs/2209.07562) [find](https://scholar.google.com/scholar?q=Xinyang+Zhang%2C+Yury+Malkov%2C+Omar+Florez%2C+Serim+Park%2C+Brian+McWilliams%2C+Jiawei+Han%2C+and+Ahmed+El%2DKishky%2E+2022%2E+Twhin%2Dbert%3A+A+socially%2Denriched+pre%2Dtrained+language+model+for+multilingual+tweet+represen)
14. Anastasia Giachanou, Bilal Ghanem, and Paolo Rosso. 2023. Detection of conspiracy propagators using psycho-linguistic characteristics. Journal of Information Science, 49(1):3–17. [find](https://scholar.google.com/scholar?q=Anastasia+Giachanou%2C+Bilal+Ghanem%2C+and+Paolo+Rosso%2E+2023%2E+Detection+of+conspiracy+propagators+using+psycho%2Dlinguistic+characteristics%2E+Journal+of+Information+Science%2C+49%281%29%3A3%E2%80%9317%2E)
15. Michael Robert Haupt, Michelle Chiu, Joseline Chang, Zoe Li, Raphael Cuomo, and Tim K Mackey. 2023. Detecting nuance in conspiracy discourse: Advancing methods in infodemiology and communication science with machine learning and qualitative content coding. Plos one, 18(12):e0295414. [find](https://scholar.google.com/scholar?q=Michael+Robert+Haupt%2C+Michelle+Chiu%2C+Joseline+Chang%2C+Zoe+Li%2C+Raphael+Cuomo%2C+and+Tim+K+Mackey%2E+2023%2E+Detecting+nuance+in+conspiracy+discourse%3A+Advancing+methods+in+infodemiology+and+communication+scien)
16. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. [Roberta: A robustly optimized BERT pretraining approach](https://md.research.elamur.ai/free/en/papers/liu-2019-roberta-robustly-optimized-bert-pretraining/). CoRR, abs/1907.11692.
17. JD Moffitt, Catherine King, and Kathleen M Carley. 2021. Hunting conspiracy theories during the covid-19 pandemic. Social Media+ Society, 7(3):20563051211043212. [find](https://scholar.google.com/scholar?q=JD+Moffitt%2C+Catherine+King%2C+and+Kathleen+M+Carley%2E+2021%2E+Hunting+conspiracy+theories+during+the+covid%2D19+pandemic%2E+Social+Media%2B+Society%2C+7%283%29%3A20563051211043212%2E)

*Updated 31 October 2026*