- Draft guidelines for publishing and citing data at NCSA.
Motivations
Why DOIs?
Discovery
Citation
Reproducibility
Transparency
NCSA Guidelines
Documentation
The underlying goal of an identifier is to allow someone to find data that served (or could serve) as evidence, correct? The underlying goal of an identifier is to allow someone to find data that served (or could serve) as evidence, correct?
Transparency
Immutability
Re: immutability. Agree that most people think of DOIs as immutable but also concur that there are several systems which allow data to be changed. If I remember right, OSF was also in this camp. A recommendation here then is if the data are not fixed, the it should be transparent when and how the data are changed. This could be done by documenting the update cycles or providing logs. We’ve likewise suggested articulation of release (and deposit) cycles for dynamic data as well.
Best effort storage
Re: best effort" storage for large datasets – that’s good phrasing. Here we’ve suggested an extra documentation file that documents how users could get access, what the minimum time commitment will be, etc. We called it a description of remote data file (see template attached), but this has never caught on.
Who assigns DOIs?
What resources should be assigned DOIs?
Timeline for DOI assignment
DOIs registered by U of I
DOIs registered by NDS
Granularity of DOI
Versioned resources
Changing or dynamic resources
Inter-institutional resources
Deprecated resources
Very large datasets
Citing datasets
Examples
See also
https://opensky.ucar.edu/islandora/object/technotes%3A503/datastream/PDF/view