HeBID: Detecting Social Identities in Hebrew-language Political Text

What it is about

HeBID is a multi-label Hebrew corpus of 5,536 sentences from Israeli politicians' Facebook posts (December 2018 to April 2021). Each sentence is marked for which of twelve social identities (for example Rightist, Leftist, Ultra-Orthodox, Zionist, Honest) it positively expresses; oppositional mentions are not labelled. The twelve were chosen by salience in a twelve-wave panel survey (N = 1,769) with a 5% threshold, so the labels follow what the public recognises. Two authors annotated, with mean Cohen's κ of 0.77. The best model is a Hebrew-tuned decoder LLM (DictaLM 2.0) at macro-F1 0.743, more than six points above the encoder models. It transfers to 500 Knesset speech excerpts at 0.72. The authors then trace prevalence, election-time surges, co-occurrence bundles and gender differences across posts, speeches and the survey.

Ideas

• I1. A positive-only rule for prag labels: a pragmatic marker fires only on positive use, and oppositional uses are recorded in a separate field. This keeps "they are corrupt" from counting as a clean-hands claim.
• I2. Choose the domain set by measurement: take a usage sample, count how often each candidate register or domain (slang, formal, humour) occurs, and keep those above a fixed threshold, as the paper does with the survey.
• I3. Measure co-occurrence bundles: count which labels co-occur in one sentence (21% of its sentences carry two or more), and compare them with the frames prag produces, as a test of multi-label output.

For Mova

I1 is a rule for prag, and it belongs with the register notes in 9. Pragmatics and politeness. I2 is a selection step for the domains of prag and the register classes of 8. Lexicon and register. I3 is a measurement for prag.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.findings-emnlp.523

@inproceedings{morlan2025hebid,
  title = {{HeBID: Detecting Social Identities in Hebrew-language Political Text}},
  author = {Mor-Lan, Guy and Rivlin-Angert, Naama and Kaplan, Yael R. and Sheafer, Tamir and Shenhav, Shaul R.},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2025},
  year = {2025},
  doi = {10.18653/v1/2025.findings-emnlp.523},
  url = {https://doi.org/10.18653/v1/2025.findings-emnlp.523}
}

References

19 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Dallas Card, Amber Boydstun, Justin Gross, Philip Resnik, and Noah Smith. 2015. The media frames corpus: Annotations of frames across issues. In Proceedings of ACL-IJCNLP 2015, pages 438–444. find
2. Dallas Card, Serina Chang, Chris Becker, Julia Mendelsohn, Rob Voigt, Leah Boustan, Ran Abramitzky, and Dan Jurafsky. 2022. Computational analysis of 140 years of us political speeches reveals more positive but increasingly polarized framing of immigration. Proceedings of the National Academy of Sciences, 119(31). find
3. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2023. Qlora: Efficient finetuning of quantized llms. Preprint, arXiv:2305.14314. ↗
4. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4171–4186. find
5. Shira Dvir-Gvirsman, Keren Tsuriel, Tamir Sheafer, Shaul Shenhav, Alon Zoizner, Liron Lavi, Michal Shamir, and Israel Waismel-Manor. 2022. Mediated representation in the age of social media: How connection with politicians contributes to citizens’ feelings of representation. evidence from a longitudinal study. Political Communication, 39(6):779–800. find
6. Gemma Team. 2024. Gemma 2: Improving open language models at a practical size. Preprint, arXiv:2408.00118. ↗
7. Noam Gidron, Lior Sheffer, and Guy Mor. 2022. The israel polarization panel dataset, 2019–2021. Electoral Studies, 80. find
8. Pere-Lluís Huguet Cabot, David Abadi, Agneta Fischer, and Ekaterina Shutova. 2021. Us vs. them: A dataset of populist attitudes, news bias and emotions. In Proceedings of the 16th Conference of the European Chapter of the ACL (EACL), pages 1921–1945. find
9. Mohit Iyyer, Peter Enns, Jordan Boyd-Graber, and Philip Resnik. 2014. Political ideology detection using recursive neural networks. In Proceedings of the 52nd Annual Meeting of the ACL, pages 1113–1122. find
10. Henri Tajfel, John Turner, William G Austin, Stephen Worchel, and 1 others. 2001. An integrative theory of intergroup conflict. Intergroup relations: Essential readings, pages 94–109. find
11. Gaël Le Mens and Aina Gallego. 2025. Positioning political texts with large language models by asking and averaging. Political Analysis, 33(3):274–282. find
12. Michael Thomas, Bo Pang, and Lillian Lee. 2006. Get out the vote: Determining support or opposition from congressional floor-debate transcripts. In Proceedings of EMNLP 2006, pages 327–335. find
13. Hauke Licht and Ronja Sczepanski. 2024. Who are they talking about? detecting mentions of social groups in political texts with supervised learning. Technical report, Working Paper. find
14. Yizhong Wang, Swaroop Mishra, Pegah Alipoormolabashi, Yeganeh Kordi, Amirreza Mirzaei, Atharva Naik, Arjun Ashok, Arut Selvan Dhanasekaran, Anjana Arunkumar, David Stap, Eshaan Pathak, Giannis Karamanolakis, Haizhi Lai, Ishan Purohit, Ishani Mondal, Jacob Anderson, Kirby Kuznia, Krima Doshi, Kuntal Kumar Pal, and 16 others. 2022. Super-NaturalInstructions: Generalization via declarative instructions on 1600+ NLP tasks. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 5085–5109, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics. find
15. Julia Mendelsohn, Ronan Le Bras, Yejin Choi, and Maarten Sap. 2023. From dogwhistles to bullhorns: Unveiling coded rhetoric with language models. In Proceedings of ACL 2023, pages 15162–15180. find
16. Guy Mor-Lan, Effi Levi, Tamir Sheafer, and Shaul R. Shenhav. 2024. IsraParlTweet: The israeli parliamentary and Twitter resource. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 9372–9381, Torino, Italia. ELRA and ICCL. find our card
17. Ines Rehbein and Josef Ruppenhofer. 2022. Who’s in, who’s out? predicting the inclusiveness or exclusiveness of personal pronouns in parliamentary debates. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 5849–5858, Marseille, France. European Language Resources Association. find
18. Sergio E. Zanotto, Qi Yu, Miriam Butt, and Diego Frassinelli. 2024. GRIT: A dataset of group reference recognition in italian. In Proceedings of LREC-COLING 2024, pages 7963–7970. find
19. Roi Zur and Ryan Bakker. 2025. The israeli parties’ positions in comparative perspective. Party Politics, 31(2):323–334. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/findings-2025-lan-hebid-detecting-social-identities-hebrew/