The README points readers at a forthcoming article:
Madore, Frédérick. "When AI Meets the Archive: Transforming the Islam West Africa Collection with Large Language Models." forthcoming.
CITATION.cff says nothing about it, so GitHub's "Cite this repository" panel offers only the software. Once the article is out, a preferred-citation block makes the panel offer the article instead, while the software entry stays available underneath — which is the right default for anyone citing this work in a paper.
What to add
preferred-citation:
type: article
title: "When AI Meets the Archive: Transforming the Islam West Africa Collection with Large Language Models"
authors:
- family-names: Madore
given-names: Frédérick
orcid: https://orcid.org/0000-0003-0959-2092
affiliation: University of Bayreuth
journal: TBD
year: TBD
doi: TBD
Blocked on
Publication details that do not exist yet — journal, year, volume/issue, pages, DOI. year and journal are required for type: article; "forthcoming" is not a valid year, so this cannot be added speculatively without breaking CFF validation again (see #8, where an incomplete identifiers block made the whole file unparseable and silently disabled the citation panel).
Also update when doing this
- The Citation section of
README.md, replacing "forthcoming" with the real reference.
preferred-citation should coexist with, not replace, the top-level software metadata.
Verify after
Validate before pushing — an invalid CFF fails silently, with no error on the repo page beyond the citation panel disappearing:
pip install cffconvert && cffconvert --validate
The README points readers at a forthcoming article:
CITATION.cffsays nothing about it, so GitHub's "Cite this repository" panel offers only the software. Once the article is out, apreferred-citationblock makes the panel offer the article instead, while the software entry stays available underneath — which is the right default for anyone citing this work in a paper.What to add
Blocked on
Publication details that do not exist yet — journal, year, volume/issue, pages, DOI.
yearandjournalare required fortype: article;"forthcoming"is not a validyear, so this cannot be added speculatively without breaking CFF validation again (see #8, where an incompleteidentifiersblock made the whole file unparseable and silently disabled the citation panel).Also update when doing this
README.md, replacing "forthcoming" with the real reference.preferred-citationshould coexist with, not replace, the top-level software metadata.Verify after
Validate before pushing — an invalid CFF fails silently, with no error on the repo page beyond the citation panel disappearing:
pip install cffconvert && cffconvert --validate