W3C home > Mailing lists > Public > www-rdf-interest@w3.org > December 2003

Re: Trust, Context, Justification and Quintuples

From: Chris Bizer <chris@bizer.de>
Date: Fri, 19 Dec 2003 11:15:26 +0100
Message-ID: <007d01c3c619$081790b0$0f8d2da0@wrz03295>
To: <www-rdf-interest@w3.org>, "Graham Klyne" <GK@ninebynine.org>

Hi Graham,

> It has been a while, but when I thought about this I came to the
conclusion
> that just one of (context-id) or (stating-id) was sufficient.
>
> E.g. given a (stating-id), (contexts) can be created as RDF containers and
> their contents can be the required collection of (stating-id).
>
> Or, if (context-id) is used, then statements or groups of statements can
be
> isolated and referenced by placing them into separate (contexts).
>

Yes sure, this is possible and classic RDF way of doing things.
The idea of quintuples as a logic model on top of RDF is making things
easier and improving query performance.
Our arguments for quintuples are:

1. A repository based on quintuples (with a relational database behind)
would have a better query performance, because statings are stored in a
compact fashion (1 row compared to 6 rows).

2. Quintuple allow to express complex queries in a compact and more human
readably way. If you try to reformulate a query like:

SELECT (null, <km:hasSkill>, <km:Programming>, ?y, null)
WHERE (?y, <to:saidBy>, ?z, null, null),
      (?z, <rdf:type>, <km:Employee>, null, null),
      (?z, <km:affiliation>, ?r, ?t, null),
      (?r, <rdf:type>, <km:Project>, null, null),
      (?r, <km:isAboutTopic>, <km:Programming>, null, null)
      (?r, <km:Duration Until>, ?s, null, null)
      (?t, <to:date>, ?u, null, null)
AND   (?s > NOW())
      (?u >"12.12.2001")
USING   km FOR <http://sampleVocabulary.org/KnowledgeManagement#>
        to FOR <http://sampleVocabulary.org/TrustOntology#>


and you are using reification and RDF containers for information modelling,
you end up with at least 3 times more patterns and joins in your database.
Beside the query would be kind of  unreadable and unmaintainable.
Robert MacGregor and  In-Young Ko argue in their paper [1] into the same
direction and gave an example comparing a triple-oriented query with a
quad-oriented query.

I think as RDF models grow, repository performance should really be an
argument.

Chris

[1]
http://km.aifb.uni-karlsruhe.de/ws/psss03/proceedings/macgregor-et-al.pdf
Received on Friday, 19 December 2003 05:18:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:03 GMT