---
title: "JTTE at ATE-IT: A CRF Model with Contextual Embeddings (2026)"
description: "The paper describes a system for Subtask A (term extraction) of the ATE-IT shared task at EVALITA 2026. 7 references, cited by 1 paper here."
authors: ["Juliette Tonneau", "Giorgio Maria Di Nunzio"]
year: 2026
venue: "Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026)"
source: "https://aclanthology.org/2026.evalita-1.51/"
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/tonneau-2026-jtte-ate-it-crf-model/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/tonneau-2026-jtte-ate-it-crf-model/)

[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/14/) / JTTE at ATE-IT: A CRF Model with Contextual Embeddings

# JTTE at ATE-IT: A CRF Model with Contextual Embeddings

**Authors:** Juliette Tonneau and [Giorgio Maria Di Nunzio](https://md.research.elamur.ai/free/en/authors/giorgio-maria-di-nunzio/) · **Year:** 2026 · **Venue:** [Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026)](https://md.research.elamur.ai/free/en/venues/evalita/) ([Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop 2026](https://md.research.elamur.ai/free/en/venues/evalita/2026/)) · **Publisher:** [CEUR Workshop Proceedings](https://md.research.elamur.ai/free/en/publishers/ceur-workshop-proceedings/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.evalita-1.51/) · [PDF](https://aclanthology.org/2026.evalita-1.51.pdf) · [OpenAlex](https://openalex.org/W7204973742)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · 1 in our library, of the 149,439 papers whose bibliographies we read

## Summary
The paper describes a system for Subtask A (term extraction) of the ATE-IT shared task at EVALITA 2026. The task is to find domain terms in Italian institutional sentences about municipal waste management. The authors treat extraction as BIO sequence tagging and use a linear-chain CRF, which keeps label sequences consistent and yields contiguous term spans.

Each token is described by three kinds of features:
- symbolic cues from spaCy, such as lemma, POS, morphology, dependency label, stopword status and neighbouring tokens;
- a frozen contextual vector from an Italian BERT model, aligned to spaCy tokens;
- a binary flag showing whether the token appears in a glossary mined automatically from the training data.

Against the official baseline, the system scores higher on span-level (Type) F1 on both the development and test sets, mainly because its precision is much better. Recall falls, however, and the gain on the test set is small, which points to limited generalization. The work shows that a light, non-fine-tuned pipeline combining classic CRF decoding with contextual embeddings is a workable, extensible baseline for Italian ATE.

## Key points
- **Toolchain:** spaCy `it_core_news_sm` handles tokenization and annotation. `dbmdz/bert-base-italian-uncased` (frozen) supplies last-layer 768-dimensional vectors, with subwords mean-pooled per spaCy token. The CRF is built with `sklearn_crfsuite`.
- **Glossary:** it starts from 50 seed lemmas (nouns and adjectives, each seen at least 5 times). Multi-word candidates are n-grams of up to six tokens, filtered by syntactic rules and required to contain a seed.
- **Type F1 vs. baseline:** 0.609 vs. 0.470 on dev and 0.498 vs. 0.469 on test.
- **Micro precision vs. baseline:** rises from 0.439 to 0.717 on dev and from 0.497 to 0.555 on test.
- **Recall on test:** micro recall falls from 0.559 to 0.448 and span-level recall from 0.508 to 0.447.
- **Proposed future work:** head and noun-phrase boundary features, multi-layer BERT feature mixing or contrastive regularization, and two-stage decoding over candidate 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

- [University of Padua](https://md.research.elamur.ai/free/en/universities/university-of-padua/) — University, Padua, Italy · 18 works · 4,260 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 · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

## Cite

Cite as: Juliette Tonneau and Giorgio Maria Di Nunzio. 2026. JTTE at ATE-IT: A CRF Model with Contextual Embeddings. Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026).

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

```bibtex
@inproceedings{tonneau2026jtte,
  title = {{JTTE at ATE-IT: A CRF Model with Contextual Embeddings}},
  author = {Juliette Tonneau and Giorgio Maria {Di Nunzio}},
  booktitle = {Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026)},
  year = {2026},
  url = {https://aclanthology.org/2026.evalita-1.51/}
}
```

- [← previous JointCoder: Exploring Automated ICD Coding on Real-World Chinese EHRs with a Multi-Agent Framework](https://md.research.elamur.ai/free/en/papers/liu-2026-jointcoder-exploring-automated-icd-coding/)
- [next → JusticeBots@LT-EDI 2026: Prompt-Based Counter-Narrative Generation for Homophobia and Transphobia Comments](https://md.research.elamur.ai/free/en/papers/pranesh-2026-justicebots-lt-edi-2026-prompt/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/14/)

## References

7 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. K. Kageura, B. Umino, Methods of automatic term recognition: A review, Terminology. International Journal of Theoretical and Applied Issues in Specialized Communication 3 (1996) 259–289. doi: 10. 1075/term.3.2.03kag . [find](https://scholar.google.com/scholar?q=K%2E+Kageura%2C+B%2E+Umino%2C+Methods+of+automatic+term+recognition%3A+A+review%2C+Terminology%2E+International+Journal+of+Theoretical+and+Applied+Issues+in+Specialized+Communication+3+%281996%29+259%E2%80%93289%2E+doi%3A+10%2E+1075)
2. G. M. Di Nunzio, S. Marchesin, G. Silvello, A systematic review of Automatic Term Extraction: What happened in 2022?, Digital Scholarship in the Humanities 38 (2023) i41–i47. doi: 10.1093/ llc/fqad030 . [find](https://scholar.google.com/scholar?q=G%2E+M%2E+Di+Nunzio%2C+S%2E+Marchesin%2C+G%2E+Silvello%2C+A+systematic+review+of+Automatic+Term+Extraction%3A+What+happened+in+2022%3F%2C+Digital+Scholarship+in+the+Humanities+38+%282023%29+i41%E2%80%93i47%2E+doi%3A+10%2E1093%2F+llc%2Ffqad030)
3. Y. Chun, M. Kim, D. Kim, C. Park, H. Lim, Enhancing Automatic Term Extraction with Large Language Models via Syntactic Retrieval, in: W. Che, J. Nabende, E. Shutova, M. T. Pilehvar (Eds.), Findings of the Association for Computational Linguistics: ACL 2025, Association for Computational Linguistics,… [find](https://scholar.google.com/scholar?q=Y%2E+Chun%2C+M%2E+Kim%2C+D%2E+Kim%2C+C%2E+Park%2C+H%2E+Lim%2C+Enhancing+Automatic+Term+Extraction+with+Large+Language+Models+via+Syntactic+Retrieval%2C+in%3A+W%2E+Che%2C+J%2E+Nabende%2C+E%2E+Shutova%2C+M%2E+T%2E+Pilehvar+%28Eds%2E%29%2C+Findings+of)
4. N. Cirillo, G. M. Di Nunzio, F. Vezzani, Ate-it at evalita 2026: Overview of the automatic term extraction italian testbed task, in: Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2026), CEUR.org, Bari, Italy, 2026. [find](https://scholar.google.com/scholar?q=N%2E+Cirillo%2C+G%2E+M%2E+Di+Nunzio%2C+F%2E+Vezzani%2C+Ate%2Dit+at+evalita+2026%3A+Overview+of+the+automatic+term+extraction+italian+testbed+task%2C+in%3A+Proceedings+of+the+Ninth+Evaluation+Campaign+of+Natural+Language+Pr)
5. F. Cutugno, A. Miaschi, A. P. Aprosio, G. Rambelli, L. Siciliani, M. A. Stranisci, Evalita 2026: Overview of the 9th evaluation campaign of natural language processing and speech tools for italian, in: Proceedings of the Ninth Evaluation Campaign of Natural Language Processing and Speech Tools for I… [find](https://scholar.google.com/scholar?q=F%2E+Cutugno%2C+A%2E+Miaschi%2C+A%2E+P%2E+Aprosio%2C+G%2E+Rambelli%2C+L%2E+Siciliani%2C+M%2E+A%2E+Stranisci%2C+Evalita+2026%3A+Overview+of+the+9th+evaluation+campaign+of+natural+language+processing+and+speech+tools+for+italian%2C+in)
6. N. Nguyen, Y. Guo, Comparisons of sequence labeling algorithms and extensions, in: Proceedings of the 24th International Conference on Machine Learning, ICML ’07, Association for Computing Machinery, New York, NY, USA, 2007, pp. 681–688. doi: 10.1145/1273496.1273582 . — [doi:10.1145/1273496.1273582](https://doi.org/10.1145/1273496.1273582) [find](https://scholar.google.com/scholar?q=N%2E+Nguyen%2C+Y%2E+Guo%2C+Comparisons+of+sequence+labeling+algorithms+and+extensions%2C+in%3A+Proceedings+of+the+24th+International+Conference+on+Machine+Learning%2C+ICML+%E2%80%9907%2C+Association+for+Computing+Machinery%2C+)
7. C. Sutton, A. McCallum, An introduction to conditional random fields, 2010. URL: [https://arxiv.org/](https://arxiv.org/) abs/1011.4088. arXiv:1011.4088 . — [arXiv:1011.4088](https://arxiv.org/abs/1011.4088) [↗](https://arxiv.org/)

*Updated 31 October 2026*