Skip to content

Latest commit

 

History

History
56 lines (35 loc) · 2.52 KB

citation.org

File metadata and controls

56 lines (35 loc) · 2.52 KB
Raw

The goal here is to create a set of standards that can be used by the Astronomical journals to enforce useful software citations. Useful here ammounts essentially to ‘readable and displayable by ADS’ (and others). The principle here is to put the responsibility for the creation and maintanance of metadata on the software authors, who have the most to gain, rather than on the paper authors who want to cite things as easily as possible.

  • The journal receives a submition of a paper by Lintott et al that used software PIPELINE by Smith et al
  • The editor uses the following policy for ensuring that there is a References entry for PIPELINE
  • The References entry may have a variety of formats (website, conference paper, Github repo) but IN ADDITION should contain a Software Reference DOI where available.
  • That DOI allows discovery (eg, through a Github release of a repository containing a PACKAGE.json or a BibTex format such as exposed by Zenodo DOI) of a set of Software Reference Metadata.
  • The Software Reference metadata will contain at minimum the following information:
    1. Title
    2. Authors
    3. Identifier
    4. Date
  • The following optional metadata:
    • Version name
    • License
  • Arbitrary additional metadata to the desire of the developers:
    • Dependencies
    • Collection tags/keywords (“Astropy”, “LSST”)
    • Description
  • ADS or any service can discover package JSON by following the link attached to the identifier. It can parse the minimum information to generate software usage and auhor citing metrics

This has several advantages which are not apparently obvious :

  1. Having the code and metadata live in a suitable open source repository will allow others to suggest adding tags for, for example,

curating sets of software, certifying that something has passed review, pointing out cool stuff. Such review is not necessary, though and can be an optional extra.

  1. With some support, this should be backwards compatible; we could pay someone or someones as a one-off to find commonly

cited code and provide references in the now-standard format

  1. Code authors are incentivised to provide good metadata by the use of that metadata for citation and its visibility on ADS
  2. The system works well for simple use cases (I’m referring to my own code in a paper) in the same way that would support a

multi-author package used by loads of people.

  1. Github will make JSON searchable soon (could be done now with careful search design) and thus this standard is inherently searchable.