Segmentation Dataset for Judgments of the Supreme Court of Justice of Portugal

The goal of this dataset is to train a segmentation model that, given a judgment from the Supreme Court of Justice of Portugal (STJ), can divide its paragraphs into sections of the judgment itself.

Dataset Contents

JSON Files:

Judgment Text:
- Contains the judgment text divided into paragraphs, with each paragraph associated with a unique ID.
Denotations:
- A list of dictionaries where each dictionary provides section information:
  - id:
    - ID of the section.
  - start:
    - Paragraph ID indicating where the section starts.
  - end:
    - Paragraph ID indicating where the section ends.
  - type:
    - Name of the section.

Possible Sections

There are 10 possible sections in the judgments:

head (cabeçalho)
report (relatório)
delimitation (delimitação)
facts (fundamentação de facto)
law (fundamentação de direito)
decision (decisão)
signature (colectivo)
declaration (declaração)
foot-notes
titles (títulos)

Variations

There is a folder that contains the original dataset. However, some sections do not always appear in the judgments. To address this, variations of the original judgments were created by deleting these absent sections (corresponding to other directories).

Contributions

Paper: Segmentation Model for Judgments of the Portuguese Supreme Court of Justice Status: Accepted but not published yet