---
title: "UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span"
description: "This paper describes the UC Santa Cruz systems for SemEval-2026 Task 10 (PsyCoMark), which studies… 0.7730 accuracy. 10 references."
authors: ["Dom Marhoefer", "Milos Suvakovic", "Glenn Grant-Richards", "Aidan Pinero", "Ryan King"]
year: 2026
venue: "Proceedings of the 20th International Workshop on Semantic Evaluation (2026)"
source: "https://aclanthology.org/2026.semeval-1.194/"
doi: "10.18653/v1/2026.semeval-1.194"
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/marhoefer-2026-ucsc-nlp-semeval-2026-task/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/marhoefer-2026-ucsc-nlp-semeval-2026-task/)

[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/54/) / UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection

# UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection

**Authors:** Dom Marhoefer, Milos Suvakovic, Glenn Grant-Richards, Aidan Pinero and Ryan King · **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-1.194/) · [PDF](https://aclanthology.org/2026.semeval-1.194.pdf) · [DOI 10.18653/v1/2026.semeval-1.194](https://doi.org/10.18653/v1/2026.semeval-1.194) · [OpenAlex](https://openalex.org/W7166862520)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
This paper describes the UC Santa Cruz systems for SemEval-2026 Task 10 (PsyCoMark), which studies conspiracy narratives in English Reddit posts. The task has two parts. The first is tagging spans for five narrative roles: Actor, Action, Effect, Evidence and Victim. The second is labeling each document as conspiratorial, not conspiratorial, or undecidable. The task is hard because critical but non-conspiratorial texts often show the same role patterns.

The team trained two independent RoBERTa-large models that share no parameters. For role extraction, every candidate span up to 32 tokens is scored for multiple roles. Each span is represented by its edge tokens, an averaged interior, a width embedding and the tokens just outside it. During training, a span counts as positive only if it nearly matches a gold span, and near-miss spans are added as hard negatives. Decoding applies per-role thresholds, containment-based NMS and span merging. For document labels, a sequence classifier with label smoothing is used.

The systems ranked 7th and 12th. The authors conclude that abstract roles are highly sensitive to boundary placement and that development metrics must mirror the official scoring.

## Key points
- **Data:** the PsyCoMark corpus has over 4,100 Reddit submission statements from more than 190 subreddits. Document labels are 35.3% Yes, 46.6% No and 18.1% Can't tell.
- **Span model training:**
  - A 6144-dimensional span vector feeds a two-layer MLP.
  - The loss is BCE with per-role positive weights, capped at 20.
  - A span is labeled positive at IoU ≥ 0.95, and hard negatives have IoU 0.50–0.75.
  - Each document contributes up to 160 sampled spans.
- **Subtask 1 results:**
  - Validation decoded micro F1 was 0.6501 (character IoU ≥ 0.3).
  - The official test gave 0.2251 macro F1 and 0.2408 micro F1 (token IoU ≥ 0.5).
  - Actor reached 0.368 test F1, while Effect reached only 0.142.
- **Subtask 2 results:**
  - Three-class validation weighted F1 was 0.6739.
  - The test gave 0.7694 weighted F1 and 0.7730 accuracy over 608 samples, because the official scorer drops "Can't tell" items.
- **Errors and limitations:** the main failure modes are boundary drift (extra modifiers included) and under-segmentation of long phrases. The 32-token span cap limits recall, and the authors propose joint modeling of the two subtasks as future work.

*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

- [University of California, Santa Cruz](https://md.research.elamur.ai/free/en/universities/university-of-california-santa-cruz/) — University, Santa Cruz, United States · 38 works · 3,604 citations

📚 Shelves (6): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,954 works · [🤝 Sociology and Political Science](https://md.research.elamur.ai/free/en/shelves/social-sciences/sociology-and-political-science/) · 626 works

## Cite

Cite as: Dom Marhoefer, Milos Suvakovic, Glenn Grant-Richards, Aidan Pinero and Ryan King. 2026. UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection. 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{marhoefer2026ucsc,
  title = {{UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection}},
  author = {Dom Marhoefer, Milos Suvakovic, Glenn Grant-Richards, Aidan Pinero and Ryan King},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.194},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.194}
}
```

- [← previous UCS: Estimating Unseen Coverage for Improved In-Context Learning](https://md.research.elamur.ai/free/en/papers/xin-2026-ucs-estimating-unseen-coverage-improved/)
- [next → uircis at SemEval-2026 Task 8: A Unified Lightweight Pipeline for Multi-Turn RAG Evaluation](https://md.research.elamur.ai/free/en/papers/zhang-2026-uircis-semeval-2026-task-8/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/54/)

## References

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

1. A George, M Ahrens, J Pierrehumbert, and M McMahon. 2024. Conspiracy detection beyond text: exploring the feasibility of adding psycho-linguistic features to enhance conspiracy detection models. In Disinformation in Open Online Media, Lecture Notes in Computer Science, pages 32–45. Springer. [find](https://scholar.google.com/scholar?q=A+George%2C+M+Ahrens%2C+J+Pierrehumbert%2C+and+M+McMahon%2E+2024%2E+Conspiracy+detection+beyond+text%3A+exploring+the+feasibility+of+adding+psycho%2Dlinguistic+features+to+enhance+conspiracy+detection+models%2E+In+Di)
2. 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)
3. Damir Korenčić, Berta Chulvi, Xavier Bonet Casals, Alejandro Toselli, Mariona Taulé, and Paolo Rosso. 2024. What distinguishes conspiracy from critical narratives? a computational analysis of oppositional discourse. Computing Research Repository. [find](https://scholar.google.com/scholar?q=Damir+Korenc%CC%8Cic%CC%81%2C+Berta+Chulvi%2C+Xavier+Bonet+Casals%2C+Alejandro+Toselli%2C+Mariona+Taul%C3%A9%2C+and+Paolo+Rosso%2E+2024%2E+What+distinguishes+conspiracy+from+critical+narratives%3F+a+computational+analysis+of+opposi)
4. Kenton Lee, Luheng He, Mike Lewis, and Luke Zettlemoyer. 2017. [End-to-end neural coreference resolution](https://md.research.elamur.ai/free/en/papers/lee-2017-end-end-neural-coreference-resolution/). In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 188–197, Copenhagen, Denmark. Association for Computational Linguistics.
5. 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/). arXiv preprint arXiv:1907.11692.
6. Zhiwei Liu, Paul Thompson, Jiaqi Rong, and Sophia Ananiadou. 2025. Conspemollm-v2: A robust and stable model to detect sentiment-transformed conspiracy theories. In ECAI 2025. IOS Press. [find](https://scholar.google.com/scholar?q=Zhiwei+Liu%2C+Paul+Thompson%2C+Jiaqi+Rong%2C+and+Sophia+Ananiadou%2E+2025%2E+Conspemollm%2Dv2%3A+A+robust+and+stable+model+to+detect+sentiment%2Dtransformed+conspiracy+theories%2E+In+ECAI+2025%2E+IOS+Press%2E)
7. 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), San Diego, United States. Association for Computational…
8. Yla R. Tausczik and James W. Pennebaker. 2010. The psychological meaning of words: Liwc and computerized text analysis methods. Journal of Language and Social Psychology, 29(1):24–54. [find](https://scholar.google.com/scholar?q=Yla+R%2E+Tausczik+and+James+W%2E+Pennebaker%2E+2010%2E+The+psychological+meaning+of+words%3A+Liwc+and+computerized+text+analysis+methods%2E+Journal+of+Language+and+Social+Psychology%2C+29%281%29%3A24%E2%80%9354%2E)
9. Jan-Willem van Prooijen and Karen M. Douglas. 2017. Conspiracy theories as part of history: The role of societal crisis situations. Memory Studies, 10(3):323– 333. [find](https://scholar.google.com/scholar?q=Jan%2DWillem+van+Prooijen+and+Karen+M%2E+Douglas%2E+2017%2E+Conspiracy+theories+as+part+of+history%3A+The+role+of+societal+crisis+situations%2E+Memory+Studies%2C+10%283%29%3A323%E2%80%93+333%2E)
10. Michael J. Wood, Karen M. Douglas, and Robbie M. Sutton. 2012. Dead and alive: Beliefs in contradictory conspiracy theories. Social Psychological and Personality Science, 3(6):767–773. [find](https://scholar.google.com/scholar?q=Michael+J%2E+Wood%2C+Karen+M%2E+Douglas%2C+and+Robbie+M%2E+Sutton%2E+2012%2E+Dead+and+alive%3A+Beliefs+in+contradictory+conspiracy+theories%2E+Social+Psychological+and+Personality+Science%2C+3%286%29%3A767%E2%80%93773%2E)

*Updated 1 November 2026*