I work in the Dataverse Project community and on its repository software and would like to ensure that the community and software follow the recommendations made in “A Data Citation Roadmap for Scholarly Data Repositories” (https://doi.org/10.1101/097196), published in October 2017.
The tenth of the roadmap’s 11 recommendations is about content negotiation: “Data repositories and identifier service providers such as identifiers.org or DataCite in addition may implement content negotiation for the persistent identifier expressed as HTTP URI, returning machine readable metadata in various formats. Content negotiation is for example supported by identifiers.org and DataCite and can return metadata in XML, RDF, Bibtex and other metadata formats.”
I’m trying determine how well the Dataverse community and software follow that recommendation. Until today I was sure that repositories that use the Dataverse software and register datasets with DataCite or EZID DOIs are already following this recommendation, since those DOIs seem to have content negotiation, but I wasn’t sure about the repositories using Handles, so a colleague suggested I also ask in the PID Forum.
Is it still the case that the Handle, DOI and ARK systems do not support real content negotiation? In the article “A Data Citation Roadmap for Scholarly Data Repositories”, the section about the tenth recommendation states that DataCite DOIs do have content negotiation, and I can follow the section’s examples to test this with DOIs from Dataverse repositories. Is that not content negotiation? Or how is that not “real” content negotiation?
Is content negotiation possible with Handles? lukask’s wrote that the relevant PID authorities should be persuaded to put this on their roadmap. Does this make sense for Handles? (Is there an “authority” for Handles in the same way that DataCite is the authority for DataCite DOIs?)
Hi @juliangautier I’m also not sure what @lukask means by “real” content negotiation, but perhaps he can explain further. DataCite supports typical server-driven content negotiation via accept header. We also support “link-based” content negotiation via data.crosscite.org, which is used in the citation formatter service. Maybe this is the variation that @lukask is referring to, as it’s not “real” content negotiation in the sense that the URL used in link-based content negotiation is not the same as the resource URL, however, real content negotiation does underly the link-based service. This is true for DataCite, as well as the other RAs that particpate in the DOI citation formatter service DOI Content Negotiation.
As for handles (which is a slippery term, since DOIs are also a handle system implementation), there is a global handle registry, operated by DONA. For the the type of handle that we’re typically talking about when using that term in the case of repositories, prefix registration is done through regional MPAs such as CNRI, which maintains the https://www.handle.net registry. While many repositories (ex, those using DSpace) do register handles in handle.net, content negotiation can’t really be done at the handle server level because handle records don’t contain descriptive metadata, as DOI records do. It would be the job of the application (ie, the repository software) that a given handle resolves to support content negotiation, since that’s where the metadata lives.
Thanks so much for this very useful info and clarifications! So far it sounds to me then that it would be possible for repository software, such as the Dataverse software, to ensure content negotiation in cases where prefix registration is done through regional MPAs, and that perhaps the community should develop this functionality into the Dataverse software to better support Dataverse-based repositories using these types of handles.
Or maybe content negotiation is or should be thought of as part of the value of using DOIs instead of other types of handles, and we should simply state that if your Dataverse repository has prefix registration done through regional MPAs, content negotiation won’t be supported.
@juliangautier While it would be great if MPAs could offer content negotiation, because handle records don’t contain descriptive metadata, content negotiation at the MPA level would require some major work. I think it’s currently more realistic to take the second approach that you suggest, in viewing content negotiation as a benefit of registering DOIs (along with high-quality DOI metadata).