[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Ref-Links] DOI & Identity Of





Just throwing this out for comments, but does anyone have any take on the
desirability of interrogating a given DOI for its identity. Is it what it
purports to be? (Can one tell this purely from the metadata that might be
deposited along with the DOI. The DOI has to be assumed to be opaque - ie "dumb"
-  although where legacy identifiers are used there may be some possibility of
correlating with the metadata. Though I wouldn't use this as an argument for
legacy identifiers :-) And again even if a correlation were established this may
well tell us nothing about the resolution data elements.)

With multi-valued, multi-typed resolution (Level 2 DOI) one primitive service
might be to implement an identity (or sanity) check on a given resolution data
element. That is, if I have a DOI as

     10.1006/geno.1999.1234

does it in fact identify what 10.1006/geno.1999.1234 should identify or is it to
be read "de facto" as the correct identifier?

With the current single-valued, single-typed - ie, URL - resolution (Level 1
DOI) the only service supported is an HTTP GET. However, one way to implement
the effect of this might be to include the identifier somewhere in the text of
the page but more reasonably as a metadata field within the content header. This
does have the advantage of being readily parsed and may be useful in validating
a "ping" test - especially if there were some recommendation of best practice.
(An HTTP status of "200 (OK)" proves nothing about the contents of the page.)

Of course, it could also be argued that whatever system generates the DOI might
generate the identical identifier as an integral metadata element and so we
would be no closer to learning the intended identity. (I think we still have the
same problem even as far as certificates are concerned.)

Or is this just a blind. The DOI identifies whatever the metadata claim - fact -
but the resolution data elements (hence services) could possibly go awry -
something like printing the wrong ISBN on a book. (Assuming that there is a
clear separation between metadata and resolution data.)

Tony




------------------------------------------------------
Ref-Links maillist  -  Ref-Links@doi.org
http://www.doi.org/mailman/listinfo/ref-links