Re: FTS comments

"Michael Rys" <mrys@microsoft.com> writes:

> 1. As I mentioned in a separate mail, the use cases should encompass
> functionality that the consensus in the WG considers to be important
> for discussion for the first version more than giving a repository
> of all possible use cases in the area of information retrieval.
>
> 2. Your use case of application of SCORE to non-text conditions is
> captured in the requirements document mainly for allowing
> researchers, vendors and at some point in time the WG to add such
> functionality. I personally consider that to be an important use
> case in the future, but do not see it as an important feature for
> this release and thus having it as part of the use cases seems a bit
> premature.

Nobody is asking you to save the whole world in two days :-)

I'm just saying that there are two important fields, DB and IR.  Doing
querying on XML documents, for me, is half IR and half DB.  But the
current XQuery effort appears to be something like 0.1 IR and 0.9 DB.

If things stay this way, then XQuery will become something that is
useful for the DB community, but it won't be useful for the IR
community.  People who want to build an IR system for XML documents
won't be able to use XQuery as the standard, not even with the
currently discussed FTS features.

There was a panel discussion during the XML-IR workshop at the SIGIR
2002 in Tampere, and the impression I got was that the XQuery effort
is not currently useful for IR and that folks are eager to help out,
if only they were asked.

I'm just a lowly PhD student who should rather be working on his
thesis.  But I had the urge to go out and wave the IR flag at least a
little bit, to do what little I can.  (I'm dreaming of a
probabilistic version of XQuery but will have to settle for a
probabilistic version of XPath, and that will be quite incomplete.
But it is going to include vagueness and uncertainty as central
concepts, in all parts of XPath.  I don't even dare dreaming of a
probabilistic RDF (yet?).)


If you say that XQuery/FTS needs its first version soon, then maybe
the way to go is to ask some IR folks to make sure you're not
preventing XQuery from being developed in the general direction of IR
later on.  Then you can save the big discussion for XQuery/FTS version
2.


Thank you for listening.
-- 
A preposition is not a good thing to end a sentence with.

Received on Tuesday, 1 April 2003 08:26:35 UTC