---
title: "asetclarity at SemEval-2026 Task 6 — paper by Pascu et al. (2026)"
description: "The paper describes the asetclarity entry for Task 1 of SemEval-2026 Task 6 (CLARITY). 15 references."
authors: ["Maria-Antonia-Emanuela Pascu", "Dan Dodun-des-Perrieres", "Daniela Gifu"]
year: 2026
venue: "Proceedings of the 20th International Workshop on Semantic Evaluation (2026)"
source: "https://aclanthology.org/2026.semeval-1.146/"
doi: "10.18653/v1/2026.semeval-1.146"
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/pascu-2026-asetclarity-semeval-2026-task-6/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/pascu-2026-asetclarity-semeval-2026-task-6/)

[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/23/) / asetclarity at SemEval-2026 Task 6: An Imbalance-Aware RoBERTa Cross-Encoder for Political Response Clarity Classification

# asetclarity at SemEval-2026 Task 6: An Imbalance-Aware RoBERTa Cross-Encoder for Political Response Clarity Classification

**Authors:** Maria-Antonia-Emanuela Pascu, Dan Dodun-des-Perrieres and [Daniela Gifu](https://md.research.elamur.ai/free/en/authors/daniela-gifu/) · **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-1.146/) · [PDF](https://aclanthology.org/2026.semeval-1.146.pdf) · [DOI 10.18653/v1/2026.semeval-1.146](https://doi.org/10.18653/v1/2026.semeval-1.146) · [OpenAlex](https://openalex.org/W7166816940)

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

## Summary
The paper describes the asetclarity entry for Task 1 of SemEval-2026 Task 6 (CLARITY). The task asks systems to decide whether an answer in a political interview is a Clear Reply, Ambivalent, or a Clear Non-Reply. The authors fine-tune a RoBERTa-large cross-encoder on the English question–answer pairs. The question and answer are concatenated into one sequence, and speaker, date and URL metadata are left out on purpose so the model cannot use them as shortcuts.

Because the label distribution is skewed, the training data is enriched for the two smaller classes. Early experiments used rule-based answer edits, and the final system used synthetic QA pairs generated by GPT-4.1 and filtered by hand. Training also uses a class-weighted cross-entropy loss. Five stratified folds are trained and their predictions are combined by soft voting.

The official submission scored 0.76 macro-F1, placing 16th of 41 systems. Most errors involve separating Clear replies from Ambivalent ones. The classifier is also packaged in a small web tool that accepts typed text or audio transcribed with Whisper. The work offers a reproducible baseline and shows where clarity detection still fails.

## Key points
- **Official result:** 0.76 macro-F1 on the leaderboard (rank 16/41) and 0.710 macro-F1 on the public Task 1 test split, which has 308 instances (79 Clear, 206 Ambivalent, 23 Clear Non-Reply).
- **Ablation on the public split:**
  - plain RoBERTa: 0.58
  - with rule-based augmentation: 0.65
  - with class-weighted loss: 0.62
  - with both: 0.68
  - full system (GPT-4.1 augmentation plus 5-fold ensemble): 0.710
  - a RoBERTa bi-encoder baseline reached 0.61.
- **Training setup:** 3 epochs, batch size 8, max length 512, bfloat16, learning rate 7.42×10⁻⁶ and weight decay 0.05. Hyperparameters were chosen by Bayesian sweeps tracked in Weights and Biases.
- **Errors:** 42% of Clear answers are predicted as Ambivalent. ROC AUC is 0.756 for Ambivalent and 0.920 for Clear Non-Reply, which is the easiest class to separate.
- **Code:** training, inference and the demo app (text input plus Whisper ASR) are released at github.com/dand101/ASET_2025_CLARITY.

*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

- [Alexandru Ioan Cuza University](https://md.research.elamur.ai/free/en/universities/alexandru-ioan-cuza-university/) — University, Iași, Romania · 8 works · 402 citations
- [Romanian Academy](https://md.research.elamur.ai/free/en/labs/romanian-academy/) — Research institute, Bucharest, Romania · 9 works · 841 citations
- [Academia Română, Filiala Iași](https://md.research.elamur.ai/free/en/labs/academia-romana-filiala-iasi/) — Research institute, Iași, Romania · 3 works · 3 citations
- [Academia Oamenilor de Știință din România](https://md.research.elamur.ai/free/en/labs/academia-oamenilor-de-stiinta-din-romania/) — Research institute, Bucharest, Romania · 3 works · 1 citation

📚 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 · [🤝 General Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/general-social-sciences/) · 254 works

## Cite

Cite as: Maria-Antonia-Emanuela Pascu, Dan Dodun-des-Perrieres and Daniela Gifu. 2026. asetclarity at SemEval-2026 Task 6: An Imbalance-Aware RoBERTa Cross-Encoder for Political Response Clarity Classification. 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{pascu2026asetclarity,
  title = {{asetclarity at SemEval-2026 Task 6: An Imbalance-Aware RoBERTa Cross-Encoder for Political Response Clarity Classification}},
  author = {Maria-Antonia-Emanuela Pascu, Dan Dodun-des-Perrieres and Daniela Gifu},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.146},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.146}
}
```

- [← previous Ascending the Infinite Ladder: Benchmarking Spatial Deformation Reasoning in Vision-Language Models](https://md.research.elamur.ai/free/en/papers/zhang-2026-ascending-infinite-ladder-benchmarking-spatial/)
- [next → Ask Me Again Differently: GRAS for Measuring Bias in Vision Language Models on Gender, Race, Age, and Skin Tone](https://md.research.elamur.ai/free/en/papers/malik-2026-ask-me-again-differently-gras/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/23/)

## References

15 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. AILS NTUA. 2026. QEvasion: CLARITY task 1 public dataset. Hugging Face Datasets. Accessed 2026-01-28. [find](https://scholar.google.com/scholar?q=AILS+NTUA%2E+2026%2E+QEvasion%3A+CLARITY+task+1+public+dataset%2E+Hugging+Face+Datasets%2E+Accessed+2026%2D01%2D28%2E)
2. Peter Bull. 1994. On identifying questions, replies, and non-replies in political interviews. Journal of Language and Social Psychology, 13(2):115–131. [find](https://scholar.google.com/scholar?q=Peter+Bull%2E+1994%2E+On+identifying+questions%2C+replies%2C+and+non%2Dreplies+in+political+interviews%2E+Journal+of+Language+and+Social+Psychology%2C+13%282%29%3A115%E2%80%93131%2E)
3. Peter Bull. 2003. The analysis of equivocation in political interviews. In Glynis M. Breakwell, editor, Doing Social Psychology Research, pages 205–228. Wiley-Blackwell, Oxford, UK. [find](https://scholar.google.com/scholar?q=Peter+Bull%2E+2003%2E+The+analysis+of+equivocation+in+political+interviews%2E+In+Glynis+M%2E+Breakwell%2C+editor%2C+Doing+Social+Psychology+Research%2C+pages+205%E2%80%93228%2E+Wiley%2DBlackwell%2C+Oxford%2C+UK%2E)
4. Peter Bull and Kate Mayer. 1993. How not to answer questions in political interviews. Political Psychology, 14(4):651–666. [find](https://scholar.google.com/scholar?q=Peter+Bull+and+Kate+Mayer%2E+1993%2E+How+not+to+answer+questions+in+political+interviews%2E+Political+Psychology%2C+14%284%29%3A651%E2%80%93666%2E)
5. Daniela Gifu. 2016. Lexical semantics in text processing. contrastive diachronic studies on romanian language. PhD, Alexandru Ioan Cuza Universtiy of Iasi, Romania. [find](https://scholar.google.com/scholar?q=Daniela+Gifu%2E+2016%2E+Lexical+semantics+in+text+processing%2E+contrastive+diachronic+studies+on+romanian+language%2E+PhD%2C+Alexandru+Ioan+Cuza+Universtiy+of+Iasi%2C+Romania%2E)
6. 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.
7. Konstantinos Thomas, Giorgos Filandrianos, Maria Lymperaiou, Chrysoula Zerva, and Giorgos Stamou. 2024. “I never said that”: A dataset, taxonomy and baselines on response clarity classification. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 5204–5233, Miami, Florida… [find](https://scholar.google.com/scholar?q=Konstantinos+Thomas%2C+Giorgos+Filandrianos%2C+Maria+Lymperaiou%2C+Chrysoula+Zerva%2C+and+Giorgos+Stamou%2E+2024%2E+%E2%80%9CI+never+said+that%E2%80%9D%3A+A+dataset%2C+taxonomy+and+baselines+on+response+clarity+classification%2E+In+Fi)
8. Konstantinos Thomas, Giorgos Filandrianos, Maria Lymperaiou, Chrysoula Zerva, and Giorgos Stamou. 2026. [Semeval-2026 task 6: Clarity – unmasking political question evasions](https://md.research.elamur.ai/free/en/papers/thomas-2026-semeval-2026-task-6-clarity/). Preprint, arXiv:2603.14027.
9. Steven E. Clayman. 2001. Answers and evasions. Language in Society, 30(3):403–442. [find](https://scholar.google.com/scholar?q=Steven+E%2E+Clayman%2E+2001%2E+Answers+and+evasions%2E+Language+in+Society%2C+30%283%29%3A403%E2%80%93442%2E)
10. Rodolfo Delmonte, Rocco Tripodi, and Daniela Gîfu. 2013. Opinion and factivity analysis of italian political discourse. In Proceedings of the 4th edition of the Italian Information Retrieval Workshop (IIR 2013), pages 88–99, Pisa, Italy. CEUR-WS on-line proceedings series. [find](https://scholar.google.com/scholar?q=Rodolfo+Delmonte%2C+Rocco+Tripodi%2C+and+Daniela+G%C3%AEfu%2E+2013%2E+Opinion+and+factivity+analysis+of+italian+political+discourse%2E+In+Proceedings+of+the+4th+edition+of+the+Italian+Information+Retrieval+Workshop+)
11. 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…
12. Vlad Ermurachi and Daniela Gîfu. 2020. Uaic1860 at [semeval-2020 task 11: Detection of propaganda techniques in news articles](https://md.research.elamur.ai/free/en/papers/martino-2020-semeval-2020-task-11-detection/). In Proceedings of the Fourteenth Workshop on Semantic Evaluation, (SemEval-2020), pages 1835–1840, Barcelona, Spain. Association for Computational Linguistics.
13. Daniela Gifu. 2010. The written press discourse and symbolic violence. presidential election analysis. PhD, Alexandru Ioan Cuza Universtiy of Iasi, Romania. [find](https://scholar.google.com/scholar?q=Daniela+Gifu%2E+2010%2E+The+written+press+discourse+and+symbolic+violence%2E+presidential+election+analysis%2E+PhD%2C+Alexandru+Ioan+Cuza+Universtiy+of+Iasi%2C+Romania%2E)
14. Daniela Gifu. 2011. In Violenta simbolica in discursul electoral, Cluj-Napoca, Romania. Casa Cartii de Stiinta. [find](https://scholar.google.com/scholar?q=Daniela+Gifu%2E+2011%2E+In+Violenta+simbolica+in+discursul+electoral%2C+Cluj%2DNapoca%2C+Romania%2E+Casa+Cartii+de+Stiinta%2E)
15. Daniela Gifu. 2013. In Temeliile Turnului Babel, Bucuresti, Romania. Editura Academiei Romane. [find](https://scholar.google.com/scholar?q=Daniela+Gifu%2E+2013%2E+In+Temeliile+Turnului+Babel%2C+Bucuresti%2C+Romania%2E+Editura+Academiei+Romane%2E)

*Updated 1 November 2026*