In looking at ARK and comparing to LSID, I see the following issues:
1. We need some mechanism to globally assert that things like
http://foobar.zaf.org/ark:/12025/654xz321
http://sneezy.dopey.com/ark:/12025/654xz321
ark:/12025/654xz321
are all the same thing. From an OWL DL perspective this is a bit
tricky because we need to use sameAs, equivalentClass, or
equivalentProperty depending on what the URI is used for.
Or, we adopt a convention that within SW documents we always make the
URI be ark:/12025/654xz321 (well, perhaps we need an extra slash - or
convince the standard to make the ark look like a urn) and always
have a property(e.g. resolvesTo) associated with it that adds the
naming authority so we can resolve it. Outside SW documents, such as
in publications, we always use the full URL.
2. To address the concern that http isn't necessarily a good
transport layer for complex data, we allow that providers may opt to
provide the metadata and policy, but return a machine and person
understandable message that redirects to use the metadata instead,
which is specified to include the sort of access service information
that lsid provides for.
3. The paper doesn't mention qualifiers, but the current version of
the specification does. We should perhaps agree on some conventions
on the form of qualifiers so that we can use them to represent
versions, where appropriate.
-Alan
