W3C home > Mailing lists > Public > public-rdf-in-xhtml-tf@w3.org > June 2008

RE: geo location tutorial cut

From: Hausenblas, Michael <michael.hausenblas@joanneum.at>
Date: Sun, 29 Jun 2008 08:24:30 +0200
Message-ID: <768DACDC356ED04EA1F1130F97D29852014C6278@RZJC2EX.jr1.local>
To: "Tim Finin" <finin@cs.umbc.edu>, "Shane McCarron" <shane@aptest.com>
Cc: "public-rdf-in-xhtml-tf.w3.org" <public-rdf-in-xhtml-tf@w3.org>

Tim, Shane,
Yes, indeed, the idea of introducing some kind of coercion seems sensible. However, I guess I'd rather go for something people are used already on the Web. Imagine a customer entering 'Jul 3 20008' into a date field. What will happen is that a client-side validation routine will flag that input error and tell the user that the expected input format is YYYY-MM-DD (and subsequently propose kinda 'did you mean 2008-07-03?').
In our case - assuming the Semantic Web stuff is exposed directly to end-users - we'll likely do a SPARQL ASK to ensure the provided resource is of type foaf:Person (which we in turn know from the schema). Otherwise we might be able to propose suitable targets to the end-user by using Semantic Web search engines and indexers or performing a look-up at e.g. DBpedia (see also faviki [1]).
[1] http://www.faviki.com/
 Michael Hausenblas, MSc.
 Institute of Information Systems & Information Management
 JOANNEUM RESEARCH Forschungsgesellschaft mbH
 Steyrergasse 17, A-8010 Graz, AUSTRIA

    phone: +43-316-876-1193 (fax:-1191)  
   e-mail: michael.hausenblas@joanneum.at
      web: http://www.joanneum.at/iis/ <https://webmail.joanneum.at/exchweb/bin/redir.asp?URL=http://www.joanneum.at/iis/> 

   mobile: +43-660-7621761
      web: http://www.sw-app.org/ <https://webmail.joanneum.at/exchweb/bin/redir.asp?URL=http://www.sw-app.org/> 


From: Tim Finin [mailto:finin@cs.umbc.edu]
Sent: Sat 2008-06-28 19:29
To: Shane McCarron
Cc: Hausenblas, Michael; public-rdf-in-xhtml-tf.w3.org
Subject: Re: geo location tutorial cut

Shane McCarron wrote:
> ... Okay - I think I understand the distinction.  And to some people that
> distinction might even be meaningful.  But what about joe-myspacer?  All
> he wants to do is say joe knows some_band.  I am certain joe is going to say:
>    myspace:joe foaf:knows myspace:some_band .
> is that wrong?  I dont think it is.  It is misleading, since what he is
> really saying litereally is "joe's myspace page knows some_band's
> myspace page".  But to the great unwashed - our target audience - is
> that distinction even meaningful?  Heck, some of those people might
> think their myspace page *is* their identity!  But if it is meaningful,
> how can we help our audience to appreciate the distinction?  And more
> importantly, how can we help our audience to use the *correct* subjects
> and objects?

I've always thought that we should explore the idea of using
something like data type coercion in programming languages to map
an illegal or apparently inappropriate value into an appropriate
one.  For example, Swoogle has found many assertions of
foaf:knows where the object is a string even thought this
property holds between two foaf:Person instances.  In many cases
very good guess, or at lease a reasonable one, at what was
intended.  On discovering an ill-formed statement like

 :steveJobs foaf:knows "Bill Gates"

we might choose to transform it into:

 :steveJobs foaf:knows [a foaf:person, foaf:name "Bill Gates"]

Advice on how to coerce bad values for properties might even be
part of an ontology.  In the example above, the foaf ontology
might specify, somehow, that if you have a string value and you
expected a foaf:Person instance, you should create one and assert
that its foaf:name is that string.

I can imagine that if we need a foaf:Person instance but have
been given a URI for a Facebook page, we could come up with
several good ways to handle this, maybe transforming



  [a foaf:Person, foaf:weblog http://www.facebook.com/profile.php?id=15205266 ]

While this seems reasonably straightforward for literal values,
there are sure to be many issues in extending it to coerce URIs
and references to RDF objects.  There is also a question about
whether a system of coercion rules for RDF should be completely
systematized (as in a programming language) or exist as a set of
heuristics.  If the latter, they could furthermore be a set of
community-based heuristics or individuals ones.


 Tim Finin, Computer Science & Electrical Engineering, Univ of Maryland
 Baltimore County, 1000 Hilltop Cir, Baltimore MD 21250. finin@umbc.edu
 http://umbc.edu/~finin 410-455-3522 fax:-3969 http://ebiquity.umbc.edu <http://ebiquity.umbc.edu/> 
Received on Sunday, 29 June 2008 06:25:11 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 29 June 2008 06:25:14 GMT