Re: interoperability (was Re: isolating shapes in named graphs)

* Peter F. Patel-Schneider <pfpschneider@gmail.com> [2014-11-26 16:26-0800]
> You appear to be arguing against interoperability.

I gave a specific worst-likely-case example of using a bdpedia
identifier and asking how the screwed up class hierarchy would
reallistically affect a publisher or consumer of data.


> The DBpedia defining documents include an ontology.  Does that
> commit users of DBpedia-minted identifiers to the DBpedia ontology?
> If not, why should use of foaf:mbox commmit one to the FOAF
> ontology, or indeed any portion of the FOAF definitions?

I believe that the most practical solution is the most intuitive one;
that the use of a term commits me to the documented semantics of that
term (vs. the whole ontology). I wanted to use a couple identifiers
(dbp:Deep_Blue and dbp:Garry_Kasparov) in order to facilite re-use of
my data. It's kind of silly that in doing so, I may be asserting that
dbp:Deep_Blue is a dbp:Athlete, but

  1 my assertion implied not that Deep Blue is an athlete, but that
    Deep Blue is a dbpedia:Athlete. DBPedia's notion of "athlete"
    (really "athlete and some other stuff") differs substantially from
    most peoples'; the semantics are sufficiently screwey that no one
    would reallistically decide something important based on a dbpedia
    type arc.

  2 many identifiers have a clear meaning and are useful enough for
    providing a shared identifier.

  3 the cause of the faulty inference is clearly DBPedia; you incur
    no added liability using the DBPedia identifier for Deep Blue.

This apprently strikes you as reckless, yet the alternative of saying
that it is a legitimate use of RDF to use foaf:mbox for shoe size
leaves us with pretty much no benefit from unique identifiers or RDF.
If we define strict laws and people break them, we can tell them to
fix them. If we say that use of a defined term need not correspond
with that definition, we lose some influence and greatly reduce the
incentive for folks to use RDF.

All communication is imperfect, yet we still hold each other responsible
for it 'cause it beats the alternative.


> peter
> 
> 
> 
> 
> On 11/26/2014 02:36 PM, Eric Prud'hommeaux wrote:
> >* Peter F. Patel-Schneider <pfpschneider@gmail.com> [2014-11-26 07:20-0800]
> >>On 11/26/2014 05:46 AM, Eric Prud'hommeaux wrote:
> >>>* Peter F. Patel-Schneider <pfpschneider@gmail.com> [2014-11-26 05:11-0800]
> >>>>One usually uses an external URI, like foaf:mbox, because one wants
> >>>>interoperability of meaning.  However, I do not believe that
> >>>>complete interoperability of URI meaning should be mandated.  I also
> >>>>do not believe that complete interoperability of URI meaning is
> >>>>possible.
> >>>>
> >>>>Further, I believe that effective interoperability can be achieved
> >>>>without mandating use of defining definitions.  For example, I may
> >>>>decide that I don't want to use the "static" part of the definition
> >>>>of foaf:mbox. Interoperability should remain for most purposes.
> >>>>
> >>>>Particular commmunities can, if they want,  require stronger
> >>>>conditions on shared meaning.  Perhaps it would be possible to set
> >>>>up a community that achieves complete interoperability of meaning.
> >>>>However, I very strongly believe that "the web" cannot be such a
> >>>>community, and thus that W3C recommendations should never mandate
> >>>>it.
> >>>
> >>>It sounds like if I'm not feeling lucky, I should never consume data
> >>>from anyone with whom I've not written up some contract. What would
> >>>that contract say? "I agree to use the vocabularies according to their
> >>>documented semantics. I will not use terms if I don't understand their
> >>>semantics."
> >>
> >>I don't think that you have to have a direct contract with that
> >>other party. There could be some out-of-band information about that
> >>other party, for example that they are a participant in some
> >>community.  There could also be information in documents, such as
> >>the use of logical properties, like rdf:type or owl:imports.  Sure a
> >>particularly perverse player could use these properties differently
> >>than you expect, either intentionally or inadvertently, but using
> >>logical properties in a non-standard fashion is something that
> >>should only be done, in my view, with great trepidation.
> >>
> >>(Of course, I'm actually going against web practice here somewhat.
> >>There are many situations where logical properties are not used
> >>correctly.  Consider owl:sameAs, for example.)
> >>
> >>>>Merging data from different sources can be problematic even if the
> >>>>use of defining definitions is mandated.  Data can be incorrect,
> >>>>after all.
> >>>
> >>>I suspect you are being a bit provocative here,
> >>
> >>Not at all, I've held this position from the very beginning of my
> >>involvement with the semantic web.  I and others have forcefully
> >>argued it at various meetings, and have affected W3C recommendations
> >>thereby.
> >>
> >>>and I'm playing along
> >>>nicely. Surely we needn't jettison this fine bathwater just because
> >>>it's slightly sullied by a baby. It's quite practical to say that I
> >>>will respect, or at least not contradict, the properties of foaf:mbox
> >>>even if there's an assertion elsewhere in that ontology that the moon
> >>>is a subclass of Things made of green cheese. What's the actual
> >>>screw-case if I use <http://dbpedia.org/resource/Moon>?
> >>
> >>Well, you do commit to lots of unusual things in this case.  For
> >>starters, there are several minimum and maximum temperatures.
> >>
> >>If you commit to the DBpedia ontology you get a number of unusual
> >>consequences, like Berlin being a mountain and Lambeau Field being a
> >>city. You also commit to unusual definitions of many categories.
> >>For example, all windmills are buildings, all libraries are
> >>educational institutions, chess players are athletes, saints are
> >>clerics, baronets are british royalty, professors are scientists.
> >
> >OK, let's take a fairly pessimal case and utter
> >   dbp:Deep_Blue :wonAgainst dbp:Garry_Kasparov .
> >and this peculiarly entails that Deep Blue is a dbp:Athlete.
> >   dbp:Deep_Blue a dbp:Athlete .
> >
> >No one will have any problems consuming our little assertion unless
> >their logic somehow interacts with the wikipedia hierarchy, in which
> >case they're already marrying an ontology with known bugs.
> >
> >
> >>I care about this sort of thing because I want to use the
> >>consequences of web information in my systems.  I don't want to have
> >>to commit to too much, however.  I also want to be able to commit to
> >>some things and see when the sources that I do use have committed to
> >>other things.  I certainly don't want to have to commit to the
> >>entire web to use any of it.
> >>
> >>peter
> >

-- 
-ericP

office: +1.617.599.3509
mobile: +33.6.80.80.35.59

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.

There are subtle nuances encoded in font variation and clever layout
which can only be seen by printing this message on high-clay paper.

Received on Thursday, 27 November 2014 21:10:20 UTC