metadata

dataset_info:
  - config_name: journalistic
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: test
        num_bytes: 28294
        num_examples: 36
      - name: train
        num_bytes: 1173238858.605204
        num_examples: 1743725
      - name: valid
        num_bytes: 587400.5
        num_examples: 1000
    download_size: 1574096064
    dataset_size: 1173854553.105204
  - config_name: legal
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: test
        num_bytes: 10385
        num_examples: 37
      - name: train
        num_bytes: 146248951.73606366
        num_examples: 465416
      - name: valid
        num_bytes: 310978
        num_examples: 1000
    download_size: 178588505
    dataset_size: 146570314.73606366
  - config_name: literature
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: test
        num_bytes: 12767
        num_examples: 36
      - name: train
        num_bytes: 30082390.51579084
        num_examples: 89522
      - name: valid
        num_bytes: 337064.5
        num_examples: 1000
    download_size: 42240467
    dataset_size: 30432222.01579084
  - config_name: politics
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: valid
        num_bytes: 2743658.864027539
        num_examples: 2000
      - name: test
        num_bytes: 64499
        num_examples: 48
      - name: train
        num_bytes: 6361013.14560123
        num_examples: 4809
    download_size: 8507277
    dataset_size: 9169171.009628769
  - config_name: social_media
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_bytes: 265591193.41674715
        num_examples: 2018904
      - name: valid
        num_bytes: 263104.33127751213
        num_examples: 2000
      - name: test
        num_bytes: 6146
        num_examples: 28
    download_size: 188296125
    dataset_size: 265860443.74802467
  - config_name: web
    features:
      - name: text
        dtype: string
      - name: label
        dtype: int64
    splits:
      - name: train
        num_bytes: 274547647.613804
        num_examples: 138867
      - name: valid
        num_bytes: 3954109.2932633953
        num_examples: 2000
      - name: test
        num_bytes: 64024
        num_examples: 34
    download_size: 164325956
    dataset_size: 278565780.90706736
configs:
  - config_name: journalistic
    data_files:
      - split: train
        path: journalistic/train-*
      - split: valid
        path: journalistic/valid-*
      - split: test
        path: journalistic/test-*
  - config_name: legal
    data_files:
      - split: train
        path: legal/train-*
      - split: valid
        path: legal/valid-*
      - split: test
        path: legal/test-*
  - config_name: literature
    data_files:
      - split: train
        path: literature/train-*
      - split: valid
        path: literature/valid-*
      - split: test
        path: literature/test-*
  - config_name: politics
    data_files:
      - split: train
        path: politics/train-*
      - split: valid
        path: politics/valid-*
      - split: test
        path: politics/test-*
  - config_name: social_media
    data_files:
      - split: train
        path: social_media/train-*
      - split: valid
        path: social_media/valid-*
      - split: test
        path: social_media/test-*
  - config_name: web
    data_files:
      - split: train
        path: web/train-*
      - split: valid
        path: web/valid-*
      - split: test
        path: web/test-*

PtBrVId

The developed corpus is a composition of pre-existing datasets initially created for other NLP tasks that provide permissive licenses. The first release of the corpus is available on Huggingface.

Data Sources

The corpus consists of the following datasets:

Domain	Variety	Dataset	Original Task	# Docs	License	Silver Labeled
Literature	PT-PT	Arquivo Pessoa	-	~4k	CC	✔
		Gutenberg Project	-	6	CC	✔
		LT-Corpus	-	56	ELRA END USER	✘
	PT-BR	Brazilian Literature	Author Identification	81	CC	✘
	PT-BR	LT-Corpus	-	8	ELRA END USER	✘
Politics	PT-PT	Koehn (2005) Europarl	Machine Translation	~10k	CC	✘
Politics	PT-BR	Brazilian Senate Speeches	-	~5k	CC	✔
Journalistic	PT-PT	CETEM Público	-	1M	CC	✘
Journalistic	PT-BR	CETEM Folha	-	272k	CC	✘
Social Media	PT-PT	Ramalho (2021)	Fake News Detection	2M	MIT	✔
	PT-BR	Vargas (2022)	Hate Speech Detection	5k	CC-BY-NC-4.0	✘
	PT-BR	Cunha (2021)	Fake News Detection	2k	GPL-3.0 license	✔
Web	BOTH	Ortiz-Suarez (2020)	-	10k	CC	✔

Table 1: Data Sources

Note: The dataset "Brazilian Senate Speeches" was created by the authors of this paper, using web crawling of the Brazilian Senate website and is available in the Huggingface repository.

Annotation Schema & Data Preprocessing Pipeline

We leveraged our knowledge of the Portuguese language to identify data sources that guaranteed mono-variety documents. However, this first release lacks any kind of supervision, so we cannot guarantee that all documents are mono-variety. In the future, we plan to release a second version of the corpus with a more robust annotation schema, combining automatic and manual annotation.

To improve the quality of the corpus, we applied a preprocessing pipeline to all documents. The pipeline consists of the following steps:

Remove all NaN values.
Remove all empty documents.
Remove all duplicated documents.
Apply the clean_text library to remove non-relevant information for language identification from the documents.
Remove all documents with a length significantly more than two standard deviations from the mean length of the documents in the corpus.

The pipeline is illustrated in Figure 1.

Image Description

Figure 1: Data Pre-Processing Pipeline

Class Distribution

The class distribution of the corpus is presented in Table 2. The corpus is highly imbalanced, with the majority of the documents being from the journalistic domain. In the future, we plan to release a second version of the corpus with a more balanced distribution across the six domains. Depending on the imbalance of the textual domain, we used different strategies to perform train-validation-test splits. For the heavily imbalanced domains, we ensured a minimum of 100 documents for validation and 400 for testing. In the other domains, we applied a stratified split.

Domain	# PT-PT	# PT-BR	Stratified
Politics	6500	4894	✓
Web	7960	21592	✓
Literature	18282	2772	✓
Law	392839	5766	✕
Journalistic	1494494	354180	✓
Social Media	2013951	6222	✕

Table 2: Class Balance across the six textual domains in both varieties of Portuguese.

Future Releases & How to Contribute

We plan to release a second version of this corpus considering more textual domains and extending the scope to other Portuguese varieties. If you want to contribute to this corpus, please contact us.