Datasets:
Expanding Ukrainian legal tasks in LEXTREME + feedback request
Joel,
Thanks again for merging the Ukrainian judgment prediction subset into LEXTREME — glad it fits well.
Two things I wanted to bring up:
1. Paper feedback. I've been running fine-tuning experiments on XLM-R and Legal-XLM-R (base + large) for temporal generalization on Ukrainian court decisions — training across three epochs (pre-war, hybrid war, full-scale invasion, 428K decisions). One interesting finding: general-purpose XLM-R outperforms Legal-XLM-R by 7–9 pp, likely due to tokenizer fertility penalties on Cyrillic. The paper cites LEXTREME and MultiLegalPile extensively. Would you be open to a quick look at the draft before submission? Your perspective on the legal model behavior would be very valuable.
2. More Ukrainian tasks for LEXTREME. Beyond judgment prediction, I have several datasets on HF that could become LEXTREME subsets:
- ua-statute-retrieval — statute section retrieval benchmark (already on arXiv: 2605.17639)
- ua-court-citation-graph — citation prediction between court decisions
- ua-temporal-drift — the temporal splits from the paper above
Happy to format these to LEXTREME specs. This would give the benchmark its first Cyrillic/Ukrainian coverage beyond judgment prediction.
3. If there are any legal NLP workshops or communities where this work would be a good fit, I'd appreciate a pointer — I'm relatively new to the venue side of the field.
Best,
Volodymyr
P.S. We also have a pipeline for 17.5M Indian court decisions (Supreme Court + 25 High Courts, 1950–2026) from the AWS Open Data Registry (CC-BY-4.0). Once the HF dataset is published, Indian judgment prediction could be another LEXTREME task — a major underrepresented jurisdiction with both English and vernacular language coverage.
Hi Volodymyr,
Thanks for the message.
- Yes, happy to look at it.
- Retrieval seems out of scope for LEXTREME unfortunately. How do you represent the citation prediction task?
- NLLP at EMNLP (https://2026.emnlp.org/calls/workshops/) could be a good venue. There might be a next edition of the DMAIL at ICDM workshop (https://dmail-workshop.github.io/DMAIL2025/) as well. And lastly there is ICAIL (https://site.smu.edu.sg/icail-2026), but the conference is soon, so you would need to wait for next year.
Hi Joel,
Sorry for the delayed follow-up — been heads-down running experiments. Two things:
1. Temporal drift paper (draft)
The paper is ready for your review. This is still a working draft, but the experiments and analysis are complete. Since my last message, we've expanded it significantly with two new experimental sections:
Section 5.5 — Continual Learning. Sequential fine-tuning across the three Ukrainian epochs. Key finding: chronological CL (pre-war → hybrid → full-scale) eliminates catastrophic forgetting for general XLM-R — pre-war F1 is fully retained (+1.8 to +6.2 pp) while full-scale gains +16.5 to +19.0 pp. Reverse-chronological CL causes severe forgetting (−12.2 to −14.3 pp). Legal-XLM-R forgets in both directions, suggesting legal-domain pretraining produces temporally fragile representations.
Section 5.6 — Cross-Jurisdictional Temporal Transfer. We use Swiss Judgment Prediction as a foreign-jurisdiction source, extending Cross-X to the temporal dimension. SJP pretraining lifts all Ukrainian epochs by +3 to +10 pp, but the forward degradation magnitude is unchanged (20.3 vs 21.3 pp). Temporal drift is orthogonal to cross-jurisdictional transfer.
The paper is 17 pages, 5 experiments, 6 tables, 5 figures. Draft PDF attached below. We'd particularly value your perspective on:
- The Legal-XLM-R finding (lower absolute but more temporally uniform -- a trade-off that could potentially be exploited)
- Whether temporal splits of SJP (2000--2020) could replicate the drift pattern in Swiss data -- this would make it a bilateral cross-jurisdictional temporal study
- Venue fit -- you mentioned NLLP at EMNLP, which seems ideal
2. Citation prediction task format
The most natural LEXTREME-compatible framing would be multi-label statute prediction (MLTC): given a court decision's facts section, predict which statutes (by article number) the court will cite in its reasoning. We'd limit the label set to the top-N most frequently cited statutes to keep it tractable.
The underlying data comes from our co-citation graph (overthelex/ua-court-citation-graph, 2.3M edges from 99.5M decisions). We can extract (facts_text → cited_statutes) pairs and format them as MLTC.
An additional dimension: the cited statute set evolves across temporal epochs (new Criminal Code articles post-2014 and post-2022), so this task would naturally combine citation prediction with temporal drift.
Would MLTC with top-N statutes work for LEXTREME, or do you have a different format in mind?
Best,
Volodymyr
Draft PDF: temporal-drift-legal-nlp-draft-2026-05-23.pdf
17 pages, compiles cleanly. Numbers are from actual training runs (MLflow), not placeholders. Still a draft — happy to iterate based on your feedback.