RE: my notes on our group discussion today on RDF Query via XQuery, in WWW-2003 BOF on RDF Query&Rules

Kudos for having such a BOF, I'm personally happy as it's touching a subject I've been 
pushing on for years... (personal bias admitted ;)
In fact, the original plan, now almost lost and buried in time, was to have RDF-Query based as far as possible
on XQuery. I think this requirement still makes lot of sense (note the "as far as possible").
I agree with all what is in these notes. I would particularly encourage people to 
look at where the crucial differences could be: 
a) what are the things you can't do with XQuery (using some serialization): study of use cases. 
Possible application of the 80/20 cut.
b) secondly, the serialization problem: the power of XQuery for RDF works much better with some
particular serializations other than others. This is a very interesting subject, and could
motivate some normative work on alternative syntaxes for RDF.
c) development of extensions to XQuery to get extra power for RDF

There are other issues but these three as the most prominent three, imho. This would have the strategical
consequence of trying a reunion between the "industrial-strength database side" and the 
"semantic-web-rdfish side", which have been kinda disjoint so far.


> -----Original Message-----
> From:
> []On Behalf Of Benjamin Grosof
> Sent: Thursday, May 22, 2003 2:21 PM
> To:;;
> Subject: my notes on our group discussion today on RDF Query via XQuery,
> in WWW-2003 BOF on RDF Query&Rules
> % notes from WWW-2003 BOF discussion on relating RDF-Query to XQuery
> % by Benjamin Grosof 5/22/03
> Here are the notes I took.  Not "official", please post 
> corrections/additions to the www-rdf-rules list.
> Benj, EricP, Andy Seaborne, DanC, general consensus:  worth looking at
> how to do RDF-Q stuff in XQuery:
> EricP:  XQuery has lots of nice stuff
> Benj and various: can reduce lots of RDF-Q stuff to XQuery, front end XQuery
> engines with more RDF-specific query lang and then do some
> query/result reformulation (incl. later for optimization); even keep
> track of; graph vs. tree will matter sometimes; can infer in rule-ish
> manner using ability to return / be constructive in head of query
> Nick Gibbons:  RDF has aspects of model-theoretic semantics we need to support
> Brian McBride:  probably OK usually to ignore whether stuff is ground or not
> DanC:  want to do some inference about what's true
> Me:  e.g., use rules for RDFS stuff (e.g., subclassof)
> TimBL:  there's a FLOWER pattern (Forall Let Where) in XQuery queries
> which is often unnecessarily complex for RDF querying, and probably hurts
> ability to optimize getting at triple stores
> Andy Seaborne: want to replace the XPath part of XQuery by something
> more RDF-specific; good to leverage on XQuery aspects that support
> things like XML-Schemas
> DanC:  want to have test cases that show what XQuery can't do that
> we need for RDF-Q
> EricP and TimBL:  let's have and RDF-Path for simpler stuff, then
> another RDF thing
> TimBL:  let's also look at having a canonical XML syntax for RDF,
> lots of people want that
> Benj:  in TimBL's keynote story about lots of XML data being put into
> RDF to be shared, the expressiveness needed will probably not be
> more, or much more, than XQuery supports, so we could get a lot of
> leverage from the above; then for stuff that starts out in life as
> RDF, we do things using other more RDF-powerful query lang/engines
> TimBL:  interesting to keep track of what started out as RDF vs. as XML
> ff-up meeting:  Fri eve
> - see by Fri noon:  look on message point, EricP
> also see WIKI (shared webspace):
> ________________________________________________________________________________________________
> Prof. Benjamin Grosof
> Web Technologies for E-Commerce, Business Policies, E-Contracting, Rules, 
> XML, Agents, Semantic Web Services
> MIT Sloan School of Management, Information Technology group
> or

Received on Thursday, 22 May 2003 11:07:56 UTC