Tools and services need some kind of identifier to make them first-class citizens. In general, the possibilities are:
- A synthetic URI, constructed purely as an identifier and not generally dereferencible
- A standard URL, following a well-organized scheme and dereferencible as a structured document (e.g., XML)
- A pre-existing URL which is strongly associated with the entity. For datasets, this would be the landing page. Also, most tools have a "homepage" from which you can get to more info, downloads, examples, etc., e.g., http://www.giss.nasa.gov/tools/panoply
In the long run, the standard URL with well-organized scheme and dereferencible as a structured document is probably preferred. Curt Tilmes of the Global Change Information System has offered to be a naming authority, giving rise to URLs like http://globalchange.gov/dataset/...
This gives rise to the question of how to "finish" the URL to provide the unique identity. Should we allow for the possibility of multiple schemes, e.g.:
http://globalchange.gov/dataset/gcmd/[dif_entry_id] http://globalchange.gov/dataset/doi/[doi] ...
This would seem to fit the GCIS problem domain of needing to account for entities donated by many agencies, each potentially with their own standards. Datsaets might end up with more than one identifier but perhaps a SameAs relationship can take care of this.
Also, such a scheme may take a while to get going, which to use in the meantime?