W3C home > Mailing lists > Public > semantic-web@w3.org > November 2012

Re: rdf semantics and timelessly true

From: Bernard Vatant <bernard.vatant@mondeca.com>
Date: Thu, 15 Nov 2012 10:09:33 +0100
Message-ID: <CAK4ZFVHPrLpRptsVqUqbHfrO02xsUQNo6Ytzwj2n-MUb+bGoeg@mail.gmail.com>
To: Pat Hayes <phayes@ihmc.us>
Cc: Semantic Web <semantic-web@w3.org>
Hi all

To go along with Pat's imaginary example of marriages and divorces, let me
give a real example I've been working on lately, the Authority Files of the
Publications Office of the European Union [1]. Authorities are record data
about countries, languages, currencies, EU corporate bodies and treaties.
Those are kept in all 23 (so far) official languages of the Union, and
every entry has a time stamp. Not only things recorded, but their various
attributes (names, codes, official languages and currencies of countries
etc). So you have to store information such as (simplifying the dates ...)

Czechoslovakia was a country from 1945 to 1993,  where it was split into
Czech Republic and Slovakia.
During this timespan, its official name changed twice, once in 1960 and
once again in 1990, and this is translated in every of the above said
languages, so for example between 1990 and 1993, the official name of
Czechoslovakia in Finnish was "Tšekin ja Slovakian liittotasavalta".

We have captured all that in RDF, trying to avoid the metaphysical pits and
being pragmatic. But you can't be loose and lazy. So you have to come up
with a model where things and their names are first-class citizens, both
potentially limited in time. That's perfectly doable, and we come up with a
generic model of time-stamped attributes with allows both back-office
management of those things and their RDF publication
This is not yet public, the priority being consumption by the various EU
institutions system, which is already a big world, and this is going at
European speed, but hopefully both model and data will be public at some
point.

Representation of time in RDF is feasible, and it will have to be done more
and more with semantic technologies going mainstream, being adopted for
keeping track of enduring and changing things.
In a word, forget the "green paradise of childish love" [2]


[1] http://publications.europa.eu/
[2] http://www.poesie.net/baudel8.htm

2012/11/15 Pat Hayes <phayes@ihmc.us>

>
> On Nov 14, 2012, at 11:13 PM, Michael Brunnbauer wrote:
>
> >
> > Hello Pat,
> >
> > people should be able to save application state like the "Counter
> hasValue"
> > example in a triple store. I do not see a problem with this as long as
> this
> > data is not published.
>
> Oh, I wholeheartedly agree. The only purpose of the RDF standards is to
> define RDF as a data *interchange* language. I have always maintained that
> "private" uses of RDF should not be governed by the specification
> constraints at all. I have been assuming that we are all talking about
> public uses of RDF, not private.
>
> > Triple stores should be used for knowledge *and* data.
> > Everything else would be quite limiting and involve a big overhead saving
> > knowledge here and data there.
> >
> > As for the foaf:name example: The degree of detail is inversely
> correlated
> > to the achievable community size (http://vimeo.com/51152934). The foaf
> > ontology has such a broad audience that making time mandatory would be
> too
> > much. If I understand you right you don't think that foaf is broken but
> > that it can be done better with a smaller community size.
>
> I think it has more to do with the expectations of the community than its
> size. FOAF is a good example of a social use of RDF which uses the data "in
> the present", assuming implicitly that it is being kept up to date and is
> always about "now". This works, though it has its dangers and is not
> strictly in conformance with the RDF specs. But imagine keeping a complete
> history of maritial relationships in an RDF triple store, with the need to
> maintain things like marriage and divorce dates, changes of name that
> result, and so on. Then I think these issues of time-dependency would need
> to be mde explcit in the data formats themselves.
>
> Pat
>



Bernard Vatant
**
Vocabularies & Data Engineering
Tel :  + 33 (0)9 71 48 84 59
Skype : bernard.vatant
Blog : the wheel and the hub <http://blog.hubjects.com/>

--------------------------------------------------------
*Mondeca**          **                   *
3 cité Nollez 75018 Paris, France
www.mondeca.com
Follow us on Twitter : @mondecanews <http://twitter.com/#%21/mondecanews>
Received on Thursday, 15 November 2012 09:10:22 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:52 GMT