The 3oloum corpus of scientific titles and abstracts
This is a corpus of titles and abstracts of 147,673 scientific articles that were scraped from the following scientific journals:
- Nature: from year 1870 to 2021
- Science: from year 1960 to 2020
- Science Advances: from year 2015 to 2020
The corpus has been created for the purpose of contributing to natural language processing (NLP) projects.
There are currently 21,309,015 words in the corpus, including the text of titles and abstracts, and excluding non-ASCII strings.
Every non-ASCII character has been replaced by the string <non_ascii> to facilitate text processing.
It is being continuously updated.