Datasets

Below is information about Spanish textual data sets created with the goal of solving NLP tasks. In this case, these are collections of texts, generally enriched with annotations.

  • QuALES

    Health
    Spanish (Uruguay)
    Published in 2022
    465
    News
    question answering

  • ParMEX-2022

    Gastronomy
    Spanish (Mexico)
    Published in 2022
    10,298
    Gastronomy documents
    paraphrasing

  • DETESTS

    News
    Spanish
    Published in 2022
    5,629
    News comments
    hate detection

  • DA-VINCIS

    Social
    Spanish (Mexico)
    Published in 2022
    5,453
    Tweets
    processing events

  • LivingNER

    Health
    Spanish
    Published in 2022
    1,850
    Clinical case reports
    (named) entity recognition

  • PoliticEs

    Politics
    Spanish
    Published in 2022
    418
    Tweets
    profiling

  • EXIST-2022-ES

    Spanish
    Published in 2022
    6,226
    Tweets
    hate detection

  • OpeNER-ES-2022

    Tourism
    Spanish
    Published in 2022
    2,057
    Reviews
    sentiment analysis

  • DisTEMIST

    Health
    Spanish
    Published in 2022
    1,000
    Clinical cases
    entity linking, (named) entity recognition

  • CT–CWT–22-ES

    Spanish
    Published in 2022
    14,990
    Tweets
    fake news detection

  • RestMEX-recommendation system

    Tourism
    Spanish (Mexico)
    Published in 2022
    2,263
    TripAdvisor comments
    recommendation systems

  • SQAC

    General, News
    Spanish
    Published in 2022
    8,817
    Encyclopedia entries News
    question answering

  • MESINESP8-T

    Health
    Spanish
    Published in 2021
    3,955
    Scientific articles
    text indexing

  • ADoBo

    Spanish
    Published in 2021
    Laws
    text classification

  • NewsCom-TOX

    Spanish
    Published in 2021
    4,359
    News comments
    hate detection

If you have published a result better than those on the list, send a message to odesia-comunicacion@lsi.uned.es indicating the result and the DOI of the article, along with a copy of it if it is not published openly.