Elamur Research⌕
☰
home / papers / 2026 / Entity-aware Cross-lingual Claim Detection for Automated Fact-checking

Entity-aware Cross-lingual Claim Detection for Automated Fact-checking

Paper · 2026 · Rrubaa Panchendrarajan and Arkaitz Zubiaga

Summary

Detecting statements that can be fact-checked is the first step of automated fact-checking. Most existing systems handle a single language or fine-tune multilingual models such as mBERT or XLM-R without testing whether they work on languages absent from training. The authors propose EX-Claim, a cross-lingual classifier for verifiable claims. It is built on the idea that a claim's verifiability depends on the entities it mentions, their types, and how well known they are.

The model combines three components:

  • XLM-R word representations;
  • entity types tagged by a retrained MultiNERD model;
  • a popularity signal, where an entity counts as popular if mGENRE can confidently link it to Wikipedia.

Entity types are learned as dense embeddings and passed through a projection layer and a simplified self-attention layer. Training data consists of COVID-19 tweets. Evaluation uses three datasets from different platforms covering 27 languages, including machine-translated synthetic data.

The entity-aware variants consistently beat a baseline without entity information, and transfer better to unseen languages. Gains are especially notable for Asian regional languages. This is relevant to low-resource multilingual fact-checking.

Key points

  • Variants: EXN-Claim uses 16 entity categories (15 MultiNERD types plus non-entity). EXP-Claim adds popularity, for 31 categories. X-Claim is the ablated version with no entity information.
  • Overall F1: EXP-Claim 0.746, EXN-Claim 0.745, X-Claim 0.735. Representing entities as one-hot vectors instead of embeddings costs about 1.1–1.3%.
  • Ablations on X-Claim: removing the projection layer drops F1 to 0.688, and removing attention drops it to 0.714.
  • Data and evaluation: datasets include CheckThat! 2022 and Kazemi 2021 (WhatsApp tiplines). The Wikipedia linking threshold is a log-probability of −0.15. Entity-aware models show lower attention entropy.
  • Limitations and release: training is limited to one topic (COVID-19) and one platform (Twitter), the method depends on NER/EL tools, and Wikipedia may introduce bias. Code and synthetic data are released on GitHub (RubaP/Ex-Claim).

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

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksSociology and Political Science · 577 works

🧭 Topics 1Research subjects and languages the work is about.Multilingual & cross-lingual NLP · 828 works

Cite

Cite as Rrubaa Panchendrarajan and Arkaitz Zubiaga. 2026. Entity-aware Cross-lingual Claim Detection for Automated Fact-checking. Findings of the Association for Computational Linguistics: EACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-eacl.2

@inproceedings{panchendrarajan2026entity,
  title = {{Entity-aware Cross-lingual Claim Detection for Automated Fact-checking}},
  author = {Rrubaa Panchendrarajan and Arkaitz Zubiaga},
  booktitle = {Findings of the Association for Computational Linguistics: EACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-eacl.2},
  url = {https://doi.org/10.18653/v1/2026.findings-eacl.2}
}

References

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

  1. S Agrestia, AS Hashemianb, and MJ Carmanc. 2022. Polimi-flatearthers at checkthat! 2022: Gpt-3 applied to claim detection. Working Notes of CLEF. find
  2. Firoj Alam, Fahim Dalvi, Shaden Shaar, Nadir Durrani, Hamdy Mubarak, Alex Nikolov, Giovanni Da San Martino, Ahmed Abdelali, Hassan Sajjad, Kareem Darwish, et al. 2021a. Fighting the covid-19 infodemic in social media: a holistic perspective and a call to arms. In Proceedings of the International AAA… find
  3. Firoj Alam, Shaden Shaar, Fahim Dalvi, Hassan Sajjad, Alex Nikolov, Hamdy Mubarak, Giovanni Da San Martino, Ahmed Abdelali, Nadir Durrani, Kareem Darwish, et al. 2021b. Fighting the covid-19 infodemic: Modeling the perspective of journalists, fact-checkers, social media platforms, policy makers, and… find
  4. Ipek Baris-Schlicht, Angel Felipe Magnossao de Paula, and Paolo Rosso. 2021. Upv at checkthat! 2021: Mitigating cultural differences for identifying multilingual check-worthy claims. In Proceedings of the Working Notes of CLEF 2021, Conference and Labs of the Evaluation Forum, Bucharest, Romania, Se… find
  5. This project is funded by the European Union and UK Research and Innovation under Grant No. 101073351 as part of Marie Skłodowska-Curie Actions (MSCA Hybrid Intelligence to monitor, promote, and analyze transformations in good democracy practices). We acknowledge Queen Mary’s Apocrita HPC facility,… — BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
  6. Włodzimierz Lewoniewski, Piotr Stolarski, Milena Stróżyna, Elzbieta Lewańska, Aleksandra Wojewoda, Ewelina Ksi˛eżniak, and Marcin Sawiński. 2024. Openfact at checkthat! 2024: Combining multiple attack methods for effective adversarial text generation. find
  7. SM Du, Sujatha Das Gollapalli, and See-Kiong Ng. 2022. Nus-ids at checkthat! 2022: identifying check-worthiness of tweets using checkthat5. Working Notes of CLEF. find
  8. Yufeng Li, Rrubaa Panchendrarajan, and Arkaitz Zubiaga. 2024. Factfinders at checkthat! 2024: Refining check-worthy statement detection with llms through data pruning. find
  9. Ahmet Bahadir Eyuboglu, Bahadir Altun, Mustafa Bora Arslan, Ekrem Sonmezer, and Mucahid Kutlu. 2023. Fight against misinformation on social media: Detecting attention-worthy and harmful tweets and verifiable and check-worthy claims. In International Conference of the Cross-Language Evaluation Forum… find
  10. Preslav Nakov, Alberto Barrón-Cedeño, Giovanni Da San Martino, Firoj Alam, Rubén Míguez, Tommaso Caselli, Mucahid Kutlu, Wajdi Zaghouani, Chengkai Li, Shaden Shaar, et al. 2022. Overview of the clef-2022 checkthat! lab task 1 on identifying relevant claims in tweets. In 2022 Conference and Labs of t… find
  11. Maram Hasanain and Tamer Elsayed. 2022. Cross-lingual transfer learning for check-worthy claim identification over twitter. arXiv preprint arXiv:2211.05087. — [arXiv:2211.05087](https://arxiv.org/abs/2211.05087) find
  12. Rrubaa Panchendrarajan and Arkaitz Zubiaga. 2024. Claim detection for automated fact-checking: A survey on monolingual, multilingual and cross-lingual research. Natural Language Processing Journal, 7:100066. find
  13. Wassim Henia, Oumayma Rjab, Hatem Haddad, and Chayma Fourati. 2021. icompass at nlp4if-2021– fighting the covid-19 infodemic. In Proceedings of the Fourth Workshop on NLP for Internet Freedom: Censorship, Disinformation, and Propaganda, pages 115–118. find
  14. Subhadarshi Panda and Sarah Ita Levitan. 2021. Detecting multilingual covid-19 misinformation on social media via contextualized embeddings. In Proceedings of the Fourth Workshop on NLP for Internet Freedom: Censorship, Disinformation, and Propaganda, pages 125–129.
  15. Ahmad Hussein, Nada Ghneim, and Ammar Joukhadar. 2021. Damascusteam at nlp4if2021: Fighting the arabic covid-19 infodemic on twitter using arabert. In Proceedings of the Fourth Workshop on NLP for Internet Freedom: Censorship, Disinformation, and Propaganda, pages 93–98.
  16. Acharya Ashish Prabhakar, Salar Mohtaj, and Sebastian Möller. 2020. Claim extraction from text using transfer learning. In Proceedings of the 17th International Conference on Natural Language Processing (ICON), pages 297–302. find
  17. Zehra Melce Hüsünbeyi, Oliver Deck, and Tatjana Scheffler. 2022. Rub-dfl at checkthat! 2022: Transformer models and linguistic features for identifying relevant. Working Notes of CLEF. find
  18. Peter J. Rousseeuw. 1987. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20:53–65. find
  19. Marcin Junczys-Dowmunt. 2019. Microsoft translator at wmt 2019: Towards large-scale document-level neural machine translation. In Proceedings of the Fourth Conference on Machine Translation (Volume 2: Shared Task Papers, Day 1), pages 225–233.
  20. A Savchev. 2022. Ai rational at checkthat! 2022: using transformer models for tweet classification. Working Notes of CLEF. find
  21. Katikapalli Subramanyam Kalyan, Ajit Rajasekharan, and Sivanesan Sangeetha. 2021. Ammus: A survey of transformer-based pretrained models in natural language processing. arXiv preprint arXiv:2108.05542. — [arXiv:2108.05542](https://arxiv.org/abs/2108.05542) find
  22. Marcin Sawiński, Krzysztof W˛ecel, Ewelina Ksi˛eżniak, Milena Stróżyna, Włodzimierz Lewoniewski, Piotr Stolarski, and Witold Abramowicz. 2023. Openfact at checkthat! 2023: Head-to-head gpt vs. bert-a comparative study of transformers language models for the detection of check-worthy. find
  23. Ashkan Kazemi, Kiran Garimella, Devin Gaffney, and Scott Hale. 2021. Claim matching beyond english to scale global fact-checking. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Vol… find
  24. Kristina P Sinaga and Miin-Shen Yang. 2020. Unsupervised k-means clustering algorithm. IEEE access, 8:80716–80727. find
  25. Thomas King, Simon Butcher, and Lukasz Zalewski. 2017. Apocrita - High Performance Computing Cluster for Queen Mary University of London. find
  26. PK Manan Suri and S Dudeja. 2022. Asatya at checkthat! 2022: multimodal bert for identifying claims in tweets. Working Notes of CLEF. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/panchendrarajan-2026-entity-aware-cross-lingual-claim/