CoolFace
Datasetpublic

LeandroRibeiro/NormasTCU

NormasTCU Overview NormasTCU iis a dataset for Legal Information Retrieval (LIR) in Brazilian Portuguese composed of normative documents from the Brazilian Federal Court of Accounts (Tribunal de Contas da União - TCU), along with queries and human-annotated relevance judgments. The dataset includes: 14,469 legal documents (normative acts); 46 queries; 812 judge query-document pairs derived from 3,048 human annotations with 3-level graded relevance.… See the full description on the dataset page: https://huggingface.co/datasets/LeandroRibeiro/NormasTCU.

sourceHugging Faceupdated 1d agoView on Hugging Face
5likes224downloads
Dataset Card

NormasTCU

Overview

NormasTCU iis a dataset for Legal Information Retrieval (LIR) in Brazilian Portuguese composed of normative documents from the Brazilian Federal Court of Accounts (Tribunal de Contas da União - TCU), along with queries and human-annotated relevance judgments.

The dataset includes:

  • 14,469 legal documents (normative acts);
  • 46 queries;
  • 812 judge query-document pairs derived from 3,048 human annotations with 3-level graded relevance.

Usage

To use the dataset, load que files docs.csv, query.csv, and qrel.csv:

python
import pandas as pd

docs = pd.read_csv("docs.csv")
queries = pd.read_csv("query.csv")
qrels = pd.read_csv("qrel.csv")

Dataset Structure

The NormasTCU dataset consists of three main files:

plaintext
NormasTCU/
│── doc.csv               # Corpus
│── query.csv             # Queries
│── qrel.csv              # Relevance judgments for each query
│── raw_human_eval.csv    # Anonymized relevance judgments from each human annotator (before aggregation)

Documents (`doc.csv`)

The doc.csv file contains 16,045 legal documents from the TCU's selected jurisprudence. Each row represents a legal document and includes the following fields:

Fields of a Document in the Corpus
FieldDescriptionTypeMissing values
KEYIDText--
UNIDADEBASICAAUTORASuperior authoring organizational unitText--
ORIGEMSource of the normative actText--
NUMNORMADocument numberText--
ANONORMADocument yearText--
TIPONORMADocument typeText--
TITULODocument titleText--
ASSUNTOSubject / summaryText7,191 (49.70%)
TEXTONORMAFull textHTML Text5 (0.03%)
DATAINICIOVIGENCIAEffective start dateDD/MM/YYYY--
DATAFIMVIGENCIAEffective end dateDD/MM/YYYY13,629 (94.19%)
SITUACAOExpressly revoked or in forceText--
TEXTOANEXOAnnex of the documentHTML Text13,237 (91.49%)
TEMAWhether the document is about external control or managementText3,686 (25.48%)
TAGSVCETagsText3,849 (26.60%)
NORMARELACIONADARelated documents of this datasetText9,918 (68.55%)

Queries (`query.csv`)

The query.csv file contains 46 standardized queries. Each row in the file contains the following fields:

FieldDescription
KEYUnique query identifier
TEXTQuery text

Relevance Judgments (qrel.csv)

The qrel.csv file contains relevance assessments for query-document pairs.

Each row in the file contains the following fields:

FieldDescription
QUERY_IDQuery identifier
DOC_IDDocument identifier
SCORERelevance score (0 = irrelevant, 1 = partially relevant, 2 = relevant)

Citation

If you use the NormasTCU dataset, please cite:

bibtex
@article{normastcu,
    title={{NormasTCU} — A {B}razilian {P}ortuguese {IR} Dataset and an Evaluation of {LLM}-as-a-Judge for Relevance Assessment}, 
    author={Leandro Carísio Fernandes and Marcus Vinícius Borela de Castro and Leandro dos Santos Ribeiro and Leonardo Augusto da Silva Pacheco and Edans Flávius de Oliveira Sandes},
    journal = {Information Processing \& Management},
    volume = {64},
    number = {2, Part A},
    year = {2027},
    issn = {0306-4573},
    doi={10.1016/j.ipm.2026.105151}
}

License

NormasTCU is provided under the CC-BY 4.0 license.

Contact

  • Leandro Carísio Fernandes Câmara dos Deputados, Brasília, Brazil Email: carisio@gmail.com
  • Marcos Vinícius Borela de Castro Tribunal de Contas da União (TCU), Brasília, Brazil Email: borela@tcu.gov.br
  • Leonardo Augusto da Silva Pacheco Tribunal de Contas da União (TCU), Brasília, Brazil Email: leonardo3108@gmail.com
  • Edans Flávius de Oliveira Sandes Tribunal de Contas da União (TCU), Brasília, Brazil Email: edansfs@tcu.gov.br

Acknowledgments

We would like to thank the Brazilian Federal Court of Accounts (TCU) for providing the documents and supporting this research.