Dr. Bunsen and the SWAG
Jonathan Rees
M. Scott Marshall
> Thanks to Kaitlin Thaney for the following.
>
> http://www.flickr.com/photos/kaythaney/3592177513/
Jonathan's post reminds me of an issue that is important to knowledge
sharing and has been on my mind lately: Scientists are often just as
concerned about *who* said something as they are about *what* was said.
The need to unequivocally identify a person is a requirement that comes
about when we share knowledge because we need to know who has provided
an assertion, and often, under what circumstances (i.e. with what
evidence, measurements, etc.). Such 'knowledge provenance' is becoming
increasingly important because systems are being developed that would
make use of both manually curated facts and those computationally
generated or 'mined'. This has been coming up in many different groups
and events, including the HCLS Scientific Discourse and BioRDF task
forces, myExperiment, HypER http://hyp-er.wik.is/ , and likely Sage[1].
Apparently, this topic also came up at the International Repositories
Infrastructure Workshop where Jonathan was present[0].
The similarity between the requirements for shared names, in general,
and 'people identifiers', in particular, is readily apparent: we would
like unambiguous and permanent URI's to be provided from an
authoritative and neutral source. I don't know about you but if Shared
Names offered people identifiers, it would be my preferred approach.
However, Shared Names has limited the scope to GO dbx records for the
moment.
Are there alternatives to the DIY do-it-yourself approach for those who
need people identifiers *today*? The only thing that I can think of is
WikiPeople[2] (which could create an awkward situation if someone else
with the name Michael Scott Marshall creates a page, who wants to be M.
Scott Marshall 2?). Oh wait, there's more at a Crossref blog [3],
although I don't think that sharing hypothetical information with other
scientists shouldn't require you to have an 'author number'.
One thing that I like about WikiPeople is that it puts identity in the
hands of the owners of the identity. Unfortunately, I think that a code
is required instead of a name to truly scale. Also, I suppose that the
most surefire way to ensure that an identity system doesn't get messy is
to require authentication e.g. a certificate from a Certificate
Authority that has high requirements for authentication such as
presenting a passport. Such levels of authentication are currently
required for European and Dutch grid certificates (finally, a use for
such seemingly exaggerated grid-burocracy!).
-Scott
[0]
http://maurice.vanderfeesten.name/blog/2009/03/20/international-repositories-infrastructure-workshop-persistent-identifiers/
[1]
http://blogs.bbsrc.ac.uk/index.php/2009/05/sage-has-its-time-a-large-scale-open-access-resource-for-systems-biologists/
[2]
http://proteins.wikiprofessional.org/index.php?title=%20WikiPeople&action=edit
[3]
http://www.crossref.org/crweblog/2009/04/the_buzz_around_people_identif.html
--
M. Scott Marshall (still have to get a PURL ;) )
http://staff.science.uva.nl/~marshall
http://adaptivedisclosure.org