Computing, Libraries, Tennis, India & other interests of Vikas Kamat
|The Problem with OpenURLs|| |
| The Problem with OpenURLs|
The OpenURL standard
was established to enable http-linking, even when accurate URL was unavailable.
It is a pretty cool initiative, allowing libraries, and writers to place
links to other people's work just based on the meta-data. I have no doubt that
this is the way to go even for non-bibliographic linking. IMO one day all the
search engines will provide efficient ways to deconstruct OpenURLs to take
the user to the most precise content item. However, the purpose of this blog
entry is to pin-point some of the holes in its specification.
- No author predicate: This omission
is unpardonable. When I first read the specification in year 2000 I wrote to
Prof. Herbert Van de Sompel, but apparently the spec. was frozen by then. In
the last three years nothing has been done. The standard tries to normalize
the author names by defining five tags aulast,aufirst,auinit,auinit1,
and auinitm which are quite useless from software implementation point of view. They are certainly limiting from a researcher's point of usability. Consider these:
- While referencing a fellow researcher's work the burden of deciphering
which is the first name, which is the last name, did the researcher
change the name after her marriage/divorce is just too painful for the
- What about the all those people who have only one name?! How
insensitive is that?
- No authentication predicates: I agree with the standard that GET is easier
to implement that POST, but they completely forgot to add authentication
predicates! Instead they advocate Cookie-pushing.
Most OpenURL implementers have used their own authentication (like token,password,ipnumber)
predicates, thus defeating the very purpose of standardization.
- Nightmare of the atitle predicate:
I want to meet the programmer who has implemented this predicate. It is
extremely difficult when it is easy, and quite impossible in most
mathematical and scientific content. I mean, how will you URL-encode
- No affiliation predicates: "I want Thomas Smith of Harward, not
Thomas Smith of Stanford." -- can't do that.
So, here's wishing for a review, and wide-spread deployment of the standard.
|(Comments Disabled for Now. Sorry!)||First Written: Tuesday, February 25, 2003|
Last Modified: 2/25/2003
|This is how I surf the web. Turns out
creating your own start page beats all portals, back-flipping,
personalized corporate pages, and book-marking tools.