"Verse-Wikisource" is a collection of 200,000 verses extracted from 9,000 works digitized by the Wikisource project.

Verses have been selected through the following process:
* All works categorized as poem (or as sub-categories of poem), using petscan.
* Only the texts parts labelled as "poem" with Wikisource internal markup system.
* Only the verses shorted than 21 words to remove remaining artifacts.

The dataset includes the following features:
* The individual verse.
* Its size.
* Its position in the original poem (which makes it possible to reconstruct the poem sequentially, if relevant)
* The page name in Wikisource.
* The link to the original document in Wikisource.