Skip to content

Datasets de los textos de cuentos de varios autorxs latinoamericanxs. Datasets benchmarks de distintas librerías de sentiment analysis en español sobre un corpus de borges.

karen-pal/borges

Repository files navigation

Borges Corpus

(WIP)

by Karen Palacio

Uso y licencia

Autorizo el uso de este dataset para cualquier tipo de uso, mientras se me de reconocimiento de autoría (y en lo posible se me informe - para dejar asentado aquí para qué ha sido utilizado y que sirva a otras personas a aprender sobre NLP)

Últimas actualizaciones

Introducción

No hay muchos recursos sobre PLN sobre textos en español.

Este es un intento de contribuir al ecosistema del mismo.

En este repo hay dos tipos de datasets: uno se correponde con los textos completos de cuentos de autorxs latinoamericanxs, y el otro contiene el análisis de sentimientos oración por oración de los textos. Para el armado se scrapeó la página Ciudad Seva con Selenium. Para el análisis de sentimientos se usó sentiment-analysis-spanish

En este repositorio además vas a encontrar los scripts que generan este dataset.

Contents

Available datasets so far in the datasets directory:

datasets/
├── arlt_full_texts.pkl
├── arredondo_full_texts.pkl
├── benedetti_full_texts.pkl
├── bombal_full_texts.pkl
├── borges_full_texts.pkl
├── carrington_full_texts.pkl
├── davila_full_texts.pkl
├── de_la_parra_full_texts.pkl
├── garro_full_texts.pkl
├── links/
├── lispector_full_texts.pkl
├── lyra_full_texts.pkl
├── ocampo_full_texts.pkl
└── sents_borges.pkl

In this repo you'll find:

  • datasets/<author>_full_texts.pkl: in here you will find the complete scraped texts in raw form, plus their metadata.
  • datasets/links/links_<author>.txt: urls, sources of the text data. used for the scraping part of the dataset building.
  • sents_<author>.pkl: a list of dataframes, corresponding to each of the texts, in the same order. This is done with the library sentiment-analysis-spanish (More libraries comming soon)
  • scraper.py: the script that builds the dataset with sentiment analysis sents.pkl
  • full_text_scraper.py: the script that builds <author>_full_texts.pkl given an author's name and the file with the links to scrap from.
  • links_scraper.py: script that builds a ./datasets/links/links_<author>.txt file, used in the scraping process.

TO DO

Expand vertically

Complementar con más datasets de otrxs autorxs y otras bibliotecas digitales

In doing so, do it with the following methodology:

  • for each male author, look for a woman author.

Trying to keep it balanced. Complicado teniendo en cuenta que incluso si hay un par de textos por autora, no se compara con la cantidad de textos de los hombres. Como no encuentro cantidad por autora voy a tener que hacer un esfuerzo a lo ancho ... es decir tener de cantidad de nombres.

Expaind horizontally

Would be of interest to also collect:

Years of each story, genre(s), author's nationality

find a way to automate this. even just a CLI would do.

Separate and expand genres

Hasta ahora solo estoy scrapeando cuentos. Estaría bueno tener poemas, "otros" , minicuentos.

Incorporate authors:

Mujeres

    Isabel Allende (1942)
    Marcela Serrano (1951) ...
    Gioconda Belli. ...
    Sor Juana Inés de la Cruz (1651 - 1695) ...
    Alfonsina Storni (1892 - 1938) ...
    Gabriela Mistral (1889 - 1957) ...
    Juana de Ibarbourou (1892 - 1979) ...

María Elena Walsh - Argentina:

Gabriela Mistral - Chile:

Claudia Cortalezzi - Argentina:

Alejandra D'Atri - Argentina:

Hombres

Salzano - Argentina:

Rubén Darío - Nicaragua

Horacio Quiroga

Cortazar

Macedonio Fernández

About

Datasets de los textos de cuentos de varios autorxs latinoamericanxs. Datasets benchmarks de distintas librerías de sentiment analysis en español sobre un corpus de borges.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published