Re: [Ref-Links] A missing piece: response protocols

[Mark Doyle <doyle@aps.org>]

> From: Eric Hellman <eric@hellman.net>
> Date: 1999-06-04 18:48:49 -0400

> One thing that has been largely unaddressed in the whole discussion of
> reference linking has been the nitty gritty of how a linking database can
> determine, in an automated way, whether a link is correct.

There are three steps I think.

1) Validate the metadata against a collective database (PubRef, PubMed, etc.)
2) Get S-Link-S info - valid metadata + up-to-date S-Link-S info should  
always give a proper link.
3) Validate the link.

Personally, I would like to be able to stop at 1 + 2. I don't want to have  
to check the validity of links every time a publisher updates their S-Link-S.  
I like to build links on the fly, not maintain huge databases of URLs. If  
broken links are generated in steps 1 and 2, then it really seems to be a  
problem with the other publishers'  system/metadata. However, since the  
centralized repository isn't set up yet for all fields, 3 is necessary at  
this point. For 3, I think publishers should provide a standard validator  
link that just checks the URL without actually returning the content behind  
the URL. Some kind of a small XML record for cross-checking would be  
sufficient. It is some about abusive to use the content links directly.

Two other points:

1) The centralized metadatabase should be a separate entity than the DOI resolver.
2) In theory, this centralized database could be replaced by reliable  
publisher validators. In practice, this might be too much of a burden on the  
publisher to have to do the batch matching. But it is food for thought.

> Is any one interested in developing a proposal in this area?

Yes.

Cheers,
Mark

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