nexoneAB/swedish-legal-decisions-raw-v1

Swedish Court Decisions — Svenska Domstolsavgöranden

55,096 court decisions spanning 45 years of Swedish case law, purpose-built for LLM training.

The most comprehensive open dataset of Swedish appellate court decisions available for AI development. Sourced directly from the official Swedish Courts case law database via their public REST API and preprocessed into three ready-to-use training configurations.

---

Why This Dataset

• Scale and depth: 55,096 decisions covering 1981 to present — the full sweep of modern Swedish jurisprudence
• Precedent-aware: Every record is flagged with ar_vagledande (is_precedent), enabling targeted training on landmark rulings that shape Swedish law
• Tri-modal design: Three subsets serve distinct training objectives — raw pretraining, instruction tuning, and structured information extraction
• Production-ready: Cleaned full text, consistent schemas, train/validation/test splits in Parquet format
• Open provenance: CC0 source data from Domstolsverket (Swedish Courts Administration), published on dataportal.se

---

Courts Covered

Decisions from all major Swedish appellate courts:

Court	Swedish name	Abbreviation
Supreme Court	Högsta domstolen	HD
Supreme Administrative Court	Högsta förvaltningsdomstolen	HFD
Courts of Appeal	Hovrätterna (Svea, Göta, Övre Norrland, m.fl.)	—
Administrative Courts of Appeal	Kammarrätterna	—
Land and Environment Court of Appeal	Mark- och miljööverdomstolen	MÖD
Labour Court	Arbetsdomstolen	AD
Migration Court of Appeal	Migrationsöverdomstolen	—

---

Dataset Configurations

pretrain — Continued Pre-training

Full-text decisions for domain-adaptive pretraining. Ideal for adapting a general Swedish or multilingual LLM to the legal domain.

Column	Type	Description
text	string	Cleaned full text of the decision
pub_id	string	Unique publication ID
domstol	string	Court name
datum	string	Decision date (YYYY-MM-DD)
typ	string	Decision type (e.g. DOM_ELLER_BESLUT)
ar_vagledande	bool	True if the decision is a binding precedent
malnummer	string	Case number
underratt_malnummer	string	Lower court case number (enables cross-instance tracing)
text_kalla	string	Text source: "html" or "pdf"
tecken	int32	Character count

Tip: Filter on ar_vagledande == True to extract the subset of landmark, precedent-setting rulings — essential for training models that reason about legal authority.

---

instruct — Instruction Tuning / SFT

Structured prompt–response pairs for supervised fine-tuning. Five legally meaningful task types, written in Swedish, cover the core competencies of a legal assistant.

Column	Type	Description
instruction	string	Task description in Swedish
input	string	Decision text (truncated if necessary)
output	string	Expected response
uppgift	string	Task type (see below)

Task types (uppgift):

Value	Description
sammanfattning	Concise summary of the decision
domslut	The court's operative ruling — what was decided
saken	Subject matter — what the case concerns
parter	Parties — appellant and respondent
skal	Legal reasoning — the court's rationale

This configuration is ready to drop into any SFT pipeline (e.g. TRL, LLaMA-Factory, Axolotl) with minimal preprocessing.

---

structured — Legal Information Extraction

Full text paired with JSON-structured section annotations, suitable for fine-tuning models on document parsing, section classification, and legal entity extraction.

Column	Type	Description
fulltext	string	Cleaned full text
sektioner_json	string	JSON object with identified sections
malnummer	string	Extracted case number

Sections in sektioner_json:

Key	Description
titel	Document title
avgörande_kort	Short ruling summary
överklagat_avgörande	The decision being appealed
klagande	Appellant
motpart	Respondent
saken	Subject matter
bakgrund	Background / facts
yrkanden	Claims and prayers for relief
skal	Legal reasoning
domslut	Operative ruling
skiljaktig	Dissenting opinion (if any)

---

Quick Start

from datasets import load_dataset

# Full-text pretraining corpus
ds = load_dataset("your-org/svenska-domstolsavgoranden", "pretrain")

# Precedent decisions only
precedents = ds["train"].filter(lambda x: x["ar_vagledande"])

# Instruction tuning
sft = load_dataset("your-org/svenska-domstolsavgoranden", "instruct")

# Structured extraction
structured = load_dataset("your-org/svenska-domstolsavgoranden", "structured")

---

Data Source & Coverage

Property	Details
Source	Sveriges Domstolar — Sök rättspraxis
API	https://rattspraxis.etjanst.domstol.se/openapi/puh-openapi.yaml
Open data registry	dataportal.se
Source license	CC0 (Domstolsverket)
Dataset license	CC BY-NC 4.0
Temporal coverage	Case law summaries from 1981 — present; full decisions from March 2025
Language	Swedish (sv)
Total records	55,096

---

Ethics & Privacy

• All decisions are public records published by Domstolsverket under CC0
• Personal data in published decisions is partially anonymized by the courts prior to publication
• This dataset is intended for research and model training purposes
• Users are responsible for ensuring GDPR compliance in downstream applications
• Not intended for re-identification of individuals

---

Citation

@dataset{swedish_court_decisions,
  title   = {Swedish Court Decisions — Svenska Domstolsavgöranden},
  source  = {domstol.se (Domstolsverket)},
  url     = {https://rattspraxis.etjanst.domstol.se},
  license = {CC0},
  records = {55096},
  years   = {1981--2025},
}

---

Legal Notice

This dataset contains court decisions published openly by Domstolsverket (Swedish Courts Administration) under CC0 license via their public REST API. Personal data has been anonymized by Domstolsverket prior to publication. Users are responsible for ensuring GDPR compliance in their own use cases. Not intended for re-identification of individuals.