README.md

metadata

dataset_info:
  features:
    - name: text
      dtype: string
    - name: tweet_id
      dtype: string
    - name: user_id
      dtype: string
  splits:
    - name: train
      num_bytes: 41994430875
      num_examples: 305011917
    - name: test
      num_bytes: 865592487
      num_examples: 6224733
  download_size: 26603606845
  dataset_size: 42860023362
  annotations_creators:
    - no-annotation
  language:
    - '''es'
    - en'
  language_creators:
    - other
  license: []
  multilinguality:
    - multilingual
  pretty_name: spanish-tweets
  size_categories:
    - 100M<n<1B
  source_datasets:
    - original
  tags:
    - language-modeling
  task_categories:
    - fill-mask

Dataset Card for "spanish-tweets"

Table of Contents

• Table of Contents
• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Creation
  • Curation Rationale
  • Source Data
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information
  • Contributions

Dataset Description

• Homepage: https://github.com/pysentimiento/robertuito
• Paper: RoBERTuito: a pre-trained language model for social media text in Spanish
• Leaderboard:
• Point of Contact: jmperez (at) dc.uba.ar

Dataset Summary

A big dataset of Spanish tweets* for pre-training language models (or other representations).

• in fact, it contains a small fraction of English, Portuguese, and other related languages

Supported Tasks and Leaderboards

Language Modeling

Languages

Mostly Spanish, but some Portuguese, English, and other languages.

Dataset Structure

Data Fields

• tweet_id: id of the tweet
• user_id: id of the user
• text: text from the tweet

Dataset Creation

The full process of data collection is described in the paper. Here we roughly outline the main points:

• A Spritzer collection uploaded to Archive.org dating from May 2019 was downloaded
• From this, we only kept tweets with language metadata equal to Spanish, and mark the users who posted these messages.
• Then, the tweetline from each of these marked users was downloaded.

This corpus consists of 622M tweets from around 432K users.

Citation Information

@inproceedings{perez-etal-2022-robertuito,
    title = "{R}o{BERT}uito: a pre-trained language model for social media text in {S}panish",
    author = "P{\'e}rez, Juan Manuel  and
      Furman, Dami{\'a}n Ariel  and
      Alonso Alemany, Laura  and
      Luque, Franco M.",
    booktitle = "Proceedings of the Thirteenth Language Resources and Evaluation Conference",
    month = jun,
    year = "2022",
    address = "Marseille, France",
    publisher = "European Language Resources Association",
    url = "https://aclanthology.org/2022.lrec-1.785",
    pages = "7235--7243",
    abstract = "Since BERT appeared, Transformer language models and transfer learning have become state-of-the-art for natural language processing tasks. Recently, some works geared towards pre-training specially-crafted models for particular domains, such as scientific papers, medical documents, user-generated texts, among others. These domain-specific models have been shown to improve performance significantly in most tasks; however, for languages other than English, such models are not widely available. In this work, we present RoBERTuito, a pre-trained language model for user-generated text in Spanish, trained on over 500 million tweets. Experiments on a benchmark of tasks involving user-generated text showed that RoBERTuito outperformed other pre-trained language models in Spanish. In addition to this, our model has some cross-lingual abilities, achieving top results for English-Spanish tasks of the Linguistic Code-Switching Evaluation benchmark (LinCE) and also competitive performance against monolingual models in English Twitter tasks. To facilitate further research, we make RoBERTuito publicly available at the HuggingFace model hub together with the dataset used to pre-train it.",
}

Contributions

Thanks to @github-username for adding this dataset.