Lyrics similarity - Githubissues

polifonia-project / sonar2021_demo

This repository is created for the documentation of the Polifonia demo that is going to be presented to SONAR2021

https://polifonia-project.github.io/sonar2021_demo/

2 stars 0 forks source link

Lyrics similarity #13

Closed enridaga closed 3 years ago

enridaga commented 3 years ago

To document activity on generating links between songs analysing the lyrics.

roccotrip commented 3 years ago

Suggestions on the csv format

add an ID for each lyrics
add a language field for each lyrics

enridaga commented 3 years ago

Added ID and Lang.

ID is a deterministic, numeric hash, not a sequential identifier. Is that OK @roccotrip ? (Cannot generate a serial ID with SPARQL Anything yet, see https://github.com/SPARQL-Anything/sparql.anything/issues/86)
Lang is assumed to be English but this will change once I add Shubert (very next step)

enridaga commented 3 years ago

OK, CSVs includes now all the lyrics from genius and songfacts, with ID and language.

See queries/README.md for instructions on how to reproduce.

roccotrip commented 3 years ago

Probably we will have to remove duplicates, i.e., two versions of the same song. At the moment, I am removing them post hoc when I compute lyrics similarity but I think it would be better to decide how to handle these cases.

ccolonna commented 3 years ago

Closing this for inactivity. Feel free to reopen.