W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > July to September 2004

RE: ACTION: discuss & promote union query (Was: ACTION: a replacement for 4.5 focussed on union query)

From: Rob Shearer <Rob.Shearer@networkinference.com>
Date: Tue, 24 Aug 2004 10:17:40 -0700
Message-ID: <CFE388CECDDB1E43AB1F60136BEB497302812F@rome.ad.networkinference.com>
To: "Steve Harris" <S.W.Harris@ecs.soton.ac.uk>, "RDF Data Access Working Group" <public-rdf-dawg@w3.org>

> OK, so I have an agregator (a single store containing 
> multiple RDF files,
> to be clear), now how do I query it without loosing the source
> information?

There's nothing about lack of aggregation as a core query language issue
that prevents you from returning this information.

What I keep coming back to for both SOURCE and PROVENANCE is that we've
clearly gone beyond our charter, because what we're querying is NOT RDF.
Want to query quads instead of triples? Fine, but you must acknowledge
that it's not RDF. Want to query a collection of independent RDF stores
organized in some undefined way? All right, but you're now querying the
collection itself, not the RDF that sits in it.

> I think you will find that a sustantial number of RDF stores 
> keep and use
> SOURCE information.

I am well aware of this, and I think it's significant that so many do. I
think there are much more sensible ways to expose this information than
hacking some special features into a query language.

I don't buy that just retrieving source information with a special query
language construct is good enough. I think people want to use it as
predicates in queries, I think they want to process that information as
true RDF, and I think that they want to use it in rules systems and
reasoning systems and all the rest.

I don't buy that source information is the only metadata that people are
going to want to attach to triples. In addition to knowing where
something was said, people will want to know by whom, and when, and so
on and so on.

I don't buy that people's main annotation of their RDF data should be
managed through document management. I thought much of the point of RDF
was that you could split your data into files in pretty much any way you
want. The same fact could come from many different places and it would
still all make sense.

I think trying to carve out a tiny niche for this non-RDF data is the
wrong approach.

> The queries tend to be fairly trivial ("I need to refresh all 
> the files
> written by John", "What file says that my name is '$firstname 
> $surname'")
> but they are important.
> 
> - Steve
> 
> 
Received on Tuesday, 24 August 2004 17:20:27 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:20 GMT