Source

Ground truth provided by TCDFT (Tribunal de Contas do Distrito Federal e Territórios)

Preprocessing done by:

• José Reinaldo da C. S. A. V. da S. Neto
• Leonardo Maffei

Dataset Information

• Dataset with retirement related named entity annotations in CoNLL format.
• Train-validation-test split was done in chronological order of DODFs, so as to not split entities of the same DODF document into different sets.

size:

• train set: 3860 sentences
• validation set: 828 sentences
• test set: 827 sentences

Named Entity Information

The number of named entities per class in each set (train/valid/test) is presented below:

Relevant Papers

Papers That Cite This Dataset