Dataset Card for Brazilian Portuguese Sentiment Analysis Dataset

Dataset Summary

Disclaimer: The team releasing the dataset did not write a dataset card for this dataset so this dataset card has been written by the contributors.

The Brazilian Portuguese Sentiment Analysis Dataset (BPSAD) is composed by the concatenation of 5 differents sources (Olist, B2W Digital, Buscapé, UTLC-Apps and UTLC-Movies), each one is composed by evaluation sentences classified according to the polarity (0: negative; 1: positive) and ratings (1, 2, 3, 4 and 5 stars).

This dataset requires manual download:

Download the concatenated file from dataset homepage.
Extract the file inside <path/to/manual/data>.
Load the dataset using the command:

datasets.load_dataset(
    path="lm4pt/bpsad",
    name='<polarity|rating>',
    data_dir='<path/to/manual/data>')

A detailed description about the dataset and the processing steps can be found at the dataset homepage.

Supported Tasks and Leaderboards

The dataset contains two configurations that represents the possible tasks related to sentiment analysis. The polarity classification is a binary classification problem where the sentences must be classified as positive (1) or negative (0) reviews. The rating prediction is a multiclass problem with values ranging from 1 to 5 stars.

Languages

The texts are in Brazilian Portuguese, as spoken by users of different e-commerces and Filmow social network.

Dataset Structure

Data Instances

polarity

{
    "review_text": "Bem macio e felpudo...recomendo.  Preço imbatível e entrega rápida. Compraria outro quando precisar",
    "polarity": 1
}

rating

{
    "review_text": "Bem macio e felpudo...recomendo.  Preço imbatível e entrega rápida. Compraria outro quando precisar",
    "rating": 4
}

Data Fields

polarity

review_text: a string feature with product or movie review.
polarity: an integer value that represents positive (1) or negative (0) reviews.

rating

review_text: a string feature with product or movie review.
rating: an integer value that represents the number of stars given by the reviewer. Possible values are 1, 2, 3, 4 and 5.

Data Splits

Data splits are created based on the original kfold column of each configuration, following the original authors recomendations:

train: folds 1 to 8
validation: fold 9
test: fold 10

	train	validation	test
polarity	1908937	238614	238613
rating	2228877	278608	278607

More information about sentence size and label distribution can be found in the dataset homepage.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

[More Information Needed]

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

@inproceedings{souza2021sentiment,
    author={
        Souza, Frederico Dias and
        Baptista de Oliveira e Souza Filho, João},
    booktitle={
        2021 IEEE Latin American Conference on
        Computational Intelligence (LA-CCI)}, 
    title={
        Sentiment Analysis on Brazilian Portuguese User Reviews}, 
    year={2021},
    pages={1-6},
    doi={10.1109/LA-CCI48322.2021.9769838}
}

Contributions

Thanks to @guilhermelmello and @DominguesPH for adding this dataset.