Re: Proposal: Searching XML

At 05:21 22-04-2002 -0400, Eliot Christian wrote:

>It might be useful to look at XPath in the context of XSLT.

Agreed, in general.

>As others have pointed out, a Z39.50 search really needs to
>operate on an abstract model rather than a concrete model.
>In effect, the concrete model is transformed to an abstract
>model with access points, the search operation acts on the
>transformed abstraction, and then records are retrieved
>according to some other transformation (GRS-1, MARC, etc.).

Agreed again.

>If we view transformations as the key to Z39.50, then we
>might see a role for XPath in the application of XSLT.
>In other words, a query might specify in XSLT a certain
>XPath expression to transform the concrete XML into an
>abstract model with access points. This abstract model

I strongly disagree with this. Yes, you can imagine that the server may 
apply XSLT to turn a concrete data structure into one that fits the search 
model in place. But under no circumstances should the client specify that 
transformation explicitly. That violates the paradigm that the client 
should not know about the server's content.

The client must specify, using a schema identifier, what data model should 
lend context to the search operands (this is exactly equivalent to our 
present use of attribute set IDs. It is the server's business how to match 
this against its own contents.

>On the retrieve side, an XSLT with XPath expressions is
>simply applied to make records into whatever syntax is
>desired. I suspect many Z39.50 implementors are already
>supporting the use of externally-defined stylesheets.

Yes, this is a good notion. Actually, given a known initial schema, it 
makes sense for the client to submit an XSLT stylesheet to request a 
transformation of each result set record prior to shipment. In my mind, 
this is exactly equivalent to the use of Espec-1, only more powerful, and 
standardised to boot.

I retract my half-proposal of earlier to use XPATH instead of GRS-1 tag 
paths in Espec-1. It would be simpler and more flexible to simply submit an 
XSLT schema.

(but this is, in the nature of Z39.50, completely unrelated to the proposal 
of using XPATH in lieu of USE attributes to select elements for searching.

--Sebastian
--
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101

Received on Monday, 22 April 2002 05:32:18 UTC