search

openjournals / joss

The Journal of Open Source Software
https://joss.theoj.org
MIT License
1.53k stars 185 forks source link

AAS/JOSS citation policy #546

Open dbouquin opened 5 years ago

dbouquin commented 5 years ago

Require AAS papers to cite both the JOSS paper and the software's Zenodo record on collaborative publications.

Why: PRISM example (ApJ paper; JOSS paper; Zenodo record)

For these collaborative papers in particular it seems like it could be an overt requirement that the AAS paper to cite the software archive directly.

dbouquin commented 5 years ago

This could also be a good place to discourage authors from listing preferred citations that don't point to the code directly. In this first example PRISM's readme on Zenodo specifically asks for a citation to the arXiv version of their ApJ paper and not their Zenodo record (and you can't find their Zenodo record unless you click through from the arXiv PDF to the referenced JOSS paper and then from there to Zenodo). The readme they have on GitHub requests citations to ApJ.

bsipocz commented 5 years ago

Oh, to start with I would be happy if journals would require to cite the latest reference, and not let papers being published with using a 6 years old reference rather than a more comprehensive one from a year ago. Unless of course they used the 6 year old software version, too.

dbouquin commented 5 years ago

If the up to date preferred citation was in a machine actionable format (#547) I think you could just write a script to cross reference the citation that should be used against the article's BibTeX.

arfon commented 5 years ago

/ cc @augustfly @crawfordsm - is this something that we could do?

augustfly commented 5 years ago

Is this this we talking about the bibtex/bibliographic latency problem or ensuring that the final ApJS paper cites both the JOSS article and the Zenodo entry? The former is entirely unrealistic right now in my opinion; the latter is entirely appropriate and do-able. For ensuring dual cites we just need us to do a post-acceptance check at the AAS Journals as well as lay down another checklist/documentation bullet.

FWIW it'd be nice if all the objects had all the single bond links we wish they could have. Right now starting at the Zenodo object one binds to the preprint version of the ApJS article via a DataCite relation ("isDocumentedBy"). Once there you can manually click through to the JOSS article or to the accepted ApJS article. It is unclear how this node will appear in future network graphs, but let's assume those edges will be direct relationships, e.g., arXiv is preprint/version of the ApJS article and arXiv cites JOSS. Nevertheless this thread began with the assertion that two edges on the network graph is 1 edge too far to walk with which I don't necessarily disagree.

Every time I think of a useful network graph discussion I'm going to cc @mfenner

dbouquin commented 5 years ago

I had meant the latter. Happy to talk through the former and graph implications though.