This dataset contains decisions from Supremo Tribunal Federal in Portuguese. These decisions were downloaded and parsed from https://www.stf.jus.br/. They are from the years 2011 to 2018.
Extract:
!tar xJvf rulingbr-v1.2.tar.xz
Load (Python):
docs = []
with open('rulingbr-v1.2.jsonl') as f:
for line in f:
obj = json.loads(line)
docs.append(obj)
obj = docs[0]
print('{}: {}'.format(obj['relator'], obj['ementa'][:52]))
For ease parsing of the file, each line contains a complete JSON object divided into 4 sections: "ementa", "acordao", "relatorio", "voto", "area", "relator", "classe", "extrato".
Ground-truth summary from another sections.
General description about the case.
Judge vote(s). It contains one or more votes from the judges. This document presents the critical aspects considered for the decision.
This is the final decision. It is the compilation of the votes.
Broad topic about the decision. Main areas:
It is the specific juridic instrument. It can be one of the following. They may also be combined.
The name of the judge responsible for composing the report and the first vote of the decision.
Additional information. This section is missing in near 50% of the cases.
@inproceedings{feijo2018rulingbr,
title={Rulingbr: A summarization dataset for legal texts},
author={de Vargas Feij{\'o}, Diego and Moreira, Viviane Pereira},
booktitle={International Conference on Computational Processing of the Portuguese Language},
pages={255--264},
year={2018},
organization={Springer},
url={https://link.springer.com/chapter/10.1007/978-3-319-99722-3_26}
}
9 commits
This dataset contains decisions from Supremo Tribunal Federal in Portuguese. These decisions were downloaded and parsed from https://www.stf.jus.br/. They are from the years 2011 to 2018.
Extract:
!tar xJvf rulingbr-v1.2.tar.xz
Load (Python):
docs = []
with open('rulingbr-v1.2.jsonl') as f:
for line in f:
obj = json.loads(line)
docs.append(obj)
obj = docs[0]
print('{}: {}'.format(obj['relator'], obj['ementa'][:52]))
For ease parsing of the file, each line contains a complete JSON object divided into 4 sections: "ementa", "acordao", "relatorio", "voto", "area", "relator", "classe", "extrato".
Ground-truth summary from another sections.
General description about the case.
Judge vote(s). It contains one or more votes from the judges. This document presents the critical aspects considered for the decision.
This is the final decision. It is the compilation of the votes.
Broad topic about the decision. Main areas:
It is the specific juridic instrument. It can be one of the following. They may also be combined.
The name of the judge responsible for composing the report and the first vote of the decision.
Additional information. This section is missing in near 50% of the cases.
@inproceedings{feijo2018rulingbr,
title={Rulingbr: A summarization dataset for legal texts},
author={de Vargas Feij{\'o}, Diego and Moreira, Viviane Pereira},
booktitle={International Conference on Computational Processing of the Portuguese Language},
pages={255--264},
year={2018},
organization={Springer},
url={https://link.springer.com/chapter/10.1007/978-3-319-99722-3_26}
}
9 commits