W3C home > Mailing lists > Public > public-awwsw@w3.org > January 2008

RE: What inferences should be made

From: Williams, Stuart (HP Labs, Bristol) <skw@hp.com>
Date: Tue, 15 Jan 2008 11:51:56 +0000
To: Alan Ruttenberg <alanruttenberg@gmail.com>
CC: "public-awwsw@w3.org" <public-awwsw@w3.org>
Message-ID: <9674EA156DA93A4F855379AABDA4A5C61194ABFB26@G5W0277.americas.hpqcorp.net>

FWIW I think that the advent of GRDDL and RDFa bring up a number of issues related to base URI and same document references.

    +--------+                     +---------------+
    |        |  Representation     | Client-Side   | Transformed Representation
    | origin +-------------------->+ GRDDL/RDFa    |------------------>
    | server |   Media Type: #1    | Transformation| Media Type: #2
    |        |   Base URI: #1      |               | Base URI: #2
    +--------+                     +---------------+

Pre-GRDDL/RDFa the contradiction you cite in the example below could not have arisen.

With care, I believe - though have not tried, a document author can arrange for the triples that they 'want' to come out of the transform, though that may involve specifying absolute URI in the relevant places of the document.

It seems to me that there are two dispositions one can take wrt to the authoring of an RDFa or GRDDL'able document.

1) Given an RDF graph, how does an author embed that graph in the document in such away that the transformed representation is a representation of that graph.
2) Given an RDFa/GRDDLable document, what triple happen to arise when it is transformed.

The first disposition is centred around authors intent, while the latter is centred around what an observer my attempt to glean from the document - and without care the two may diverge.

In respect of the example below: clearly http://sw-app.org/mic.xhtml#i cannot be the URI of a piece of document serialisation and the URI of a person.

IMO the HTML and RDF media types are incompatible for mixing in this way. The RDF media-type has an inherent "thing-described-by" indirection whereas the HTML media type does not. I can only see the representation and transformed representation as being consistent where either a given anchor (ID or rdf:ID) appear in only one or the RDF assertions are indeed about the relevant parts of the document.

In this case... how would the contradiction be detected?

http://sw-app.org/mic.xhtml contains the following served as:  application/xhtml+xml

                <div id="i" instanceof="foaf:Person" about="http://sw-app.org/mic.xhtml#i">
                <p>
                        My name is <span property="foaf:name" datatype="xsd:string">Michael Hausenblas</span>
                        (also known as <span property="foaf:nick">mic.</span>) and
                        I work at <a href="http://www.joanneum.at/iis/" rel="foaf:workplaceHomepage">JOANNEUM RESEARCH</a> in the field of media semantics.
                        If you are interested in further details, you may also want to visit my
                        <a rel="foaf:homepage" href="http://sw-app.org/about.html">private homepage</a>
                        or read my <a rel="foaf:homepage" href="http://mhausenblas.blogr.com">blog</a>.
                </p>

                        <div class="centerArea">
                                <img src="http://sw-app.org/img/mic_2007_01.jpg" rel="foaf:depiction" alt="Picture of Michael Hausenblas, 2007" href="http://sw-app.org/img/mic_2007_01.jpg"/>
                        </div>

                        <h2>Business Contact</h2>
                        <div rel="foaf:office">
                <span rel="foaf:address">
              <span property="foaf:street">Steyrergasse 17</span><br />
                <span property="foaf:postalCode">8020</span> <span property="foaf:city">Graz</span> <br />
                    <span property="foaf:country">Austria</span><br />
              <span property="foaf:phone">tel:+43-(316)-876-1193</span><br />
                <span property="foaf:mail" datatype="xsd:string"><a href="mailto:michael.hausenblas@joanneum.at">michael.hausenblas@joanneum.at</a></span>
                </span>
        </div>

The contradiction arises in the very first line where 'id="i"' introduces an anchor whose URI, given the document base URI is <http://sw-app.org/mic.xhtml#i>.

Implicit in the application/xhtml+xml media-type is that that URI identifies a part of the document - some background rule is needed somewhere to be able to say that ie.

        <http://sw-app.org/mic.xhtml#i> a webarch:DocumentFragment .

Ironically, maybe, the 'about="http://sw-app.org/mic.xhtml#i"' attribute in the same element which establishes the subject of the RDF assertions emitted in the transformed representation is given in full (and could have been for a different subject URI).

In concept at least, the domain constraints (assuming they are stated in the FOAF ontology - I haven't checked) on the foaf properties used license an inference that:

        <http://sw-app.org/mic.xhtml#i> a foaf:Person .

That foaf:Person and webarch:DocumentFragment are disjoint needs to be a seperately stated piece of background knowledge in order for a reasoner to detect the contradiction.

IMO it comes down to an author deciding what they *want* the URIs they create to stand for. In this case a change to the first line of either

        <div id="me" instanceof="foaf:Person" about="http://sw-app.org/mic.xhtml#i">
or
        <div id="i" instanceof="foaf:Person" about="http://sw-app.org/mic.xhtml#mic">

would avoid this particular contradiction, establishing separate URI for document fragment and person (only the latter of which propagates to the RDF - I think) - modulo other assertions which may contribute to contradictions in other ways.

A simpler Lint-like check that flags up when 'id' and 'about' attributes on the same element resolve to the same URI would pick up a great many situations where the authors intent is unclear.


BR,

Stuart
--
Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England



________________________________

        From: public-awwsw-request@w3.org [mailto:public-awwsw-request@w3.org] On Behalf Of Alan Ruttenberg
        Sent: 30 December 2007 00:06
        To: public-awwsw@w3.org
        Subject: What inferences should be made


        The contradiction noted below would be a candidate for one sort of thing we would like to detect.
        -Alan

        Begin forwarded message:


                Resent-From: semantic-web@w3.org
                From: Richard Cyganiak <richard@cyganiak.de>
                Date: December 28, 2007 8:21:45 AM EST
                To: Leo Sauermann <leo.sauermann@dfki.de>
                Cc: "Hausenblas, Michael" <michael.hausenblas@joanneum.at>, Leo Sauermann <sauermann@dfki.uni-kl.de>, semantic-web@w3.org
                Subject: Re: Cool URIs, the Semantic Web and Everything
                Archived-At: <http://www.w3.org/mid/D5C350C2-E110-4672-926D-28373DAAA4B7@cyganiak.de>



                On 25 Dec 2007, at 21:41, Leo Sauermann wrote:
                [snip]

                                        Q.II:  What does http://sw-app.org/mic.xhtml#i identfiy?

                [snip]

                                This is impossible to answer, because the URI's configuration is broken. Even the author of the document seems to be confused about what he wants the URI to identify.

                                There is an XHTML representation, and it has a id="i", which indicates that the URI identifies an XHTML fragment.

                                But the XHTML document also encodes an RDF graph using RDFa. In it, the author tries to use the same URI to denote a person. He claims that a document fragment is a person. That's a nonsensical statement.

                                Fortunately, this is easy to fix: Remove the id="i" from the document, or change it to a different ID, and everything is fine. After that fix, the answer would be 1, 2 and 6.


                        I don't know about removing the ID, maybe this would be good. But I would not make a "must" out of it, why not keep both


                The application/xhtml+xml MIME type registration says: If there is @id="i" in the XHTML document, then <mic.xhtml#i> designates a part of that document. Thus, web architecture clearly states that <mic.xhtml#i> identifies a document part.

                Michael's RDFa, however, says that <mic.xhtml#i> denotes a person.

                That's a contradiction. A person is not a section of a document. Therefore, the @id="i" *MUST* be changed or removed, otherwise we have a URI collision.

                Richard






                        best
                        Leo







                                        Cheers,
                                           Michael

                                        ----------------------------------------------------------
                                        Michael Hausenblas, MSc.
                                        Institute of Information Systems & Information Management
                                        JOANNEUM RESEARCH Forschungsgesellschaft mbH
                                        Steyrergasse 17, A-8010 Graz, AUSTRIA
                                        ----------------------------------------------------------


                                        -----Original Message-----
                                        From: Richard Cyganiak [mailto:richard@cyganiak.de]
                                        Sent: Friday, December 21, 2007 5:26 PM
                                        To: Hausenblas, Michael
                                        Cc: Leo Sauermann; semantic-web@w3.org; Leo Sauermann
                                        Subject: Re: Cool URIs, the Semantic Web and Everything

                                        Michael,

                                        On 21 Dec 2007, at 08:23, Hausenblas, Michael wrote:

                                        In Cool URIs you are
                                        referring to a certain
                                        setup ('deployment scenarios in which the RDF data and the

                                        HTML data

                                        is served separately').
                                        Also the figure right before section 3.1 suggests that there is an
                                        explicit RDF document and an HTML document, each with a

                                        distinct URL.

                                        As you know, this is not the case with RDFa.


                                        Would changing the sentence

                                        "In those cases [RDFa, microformats and GRDDL] the RDF data
                                        is extracted from the returned HTML document."

                                        to

                                        "In those cases, the RDF data is extracted from the HTML
                                        document and no separate RDF document is needed."

                                        address your complaint?

                                        The rest of the document's narrative is consistent with use
                                        of RDFa, as far as I can tell.

                                        Best,
                                        Richard





                                        So, that is were my confusion stems from. I know that due to time
                                        constraints you decided that this is the way it is. It

                                        would still be

                                        nice to learn why the figure right before section 3.1

                                        (sorry, no label

                                        available) 'shows the desired relationships between a

                                        resource and its

                                        describing documents'.

                                        Cheers,
                                           Michael

                                        [1]

                                        http://lists.w3.org/Archives/Public/semantic-web/2007Dec/0121.html


                                        ----------------------------------------------------------
                                        Michael Hausenblas, MSc.
                                        Institute of Information Systems & Information Management JOANNEUM
                                        RESEARCH Forschungsgesellschaft mbH

                                        http://www.joanneum.at/iis/
                                        ----------------------------------------------------------



                                        -----Original Message-----
                                        From: Leo Sauermann [mailto:sauermann@dfki.uni-kl.de]
                                        Sent: Thursday, December 20, 2007 12:36 PM
                                        To: Hausenblas, Michael
                                        Cc: semantic-web@w3.org; Leo Sauermann
                                        Subject: Re: Cool URIs, the Semantic Web and Everything

                                        Hausenblas, Michael schrieb:

                                           Leo,
                                               Thanks for your explanation. I remain not totally convinced :)

                                        good, then give a practical example (using concrete RDFa

                                        code) where

                                        you think some work needs to be done and provide a

                                        suggestion how to

                                        solve it. That you are not convinced may be caused by

                                        various reasons

                                        we don't know about, shine light on them.



                                               So, *if* we agree on what you said, IMHO we should
                                           reconsider the following paragraph in 'Cool URIs' [1]:
                                               'The solutions described in the following apply to deployment
                                        scenarios
                                           in which the RDF data and the HTML data is served

                                        separately, such

                                        as a
                                           standalone RDF/XML document
                                           along with an HTML document. The metadata can also be

                                        embedded in

                                        HTML,
                                           using technologies such as
                                           RDFa [RDFa Primer], microformats and other documents to

                                        which the

                                        GRDDL
                                           [GRDDL] mechanisms can be applied.
                                           In those cases the RDF data is extracted from the returned HTML
                                           document.'

                                        I see no reason for changes until you exactly specify where this
                                        paragraph contradicts http-range-14 or other TAG

                                        resolutions or W3C

                                        recommendations.

                                        the point is that RDF/XML, N3, RDFa and GRDDL are

                                        mimetypes encoding

                                        RDF triples while URIs are something used inside these RDF

                                        triples,

                                        so at the beginning both are completly different and do not affect
                                        each other.

                                        "Cool uris" is about URIs and not about RDF serialization.

                                        best
                                        Leo



                                               Still unsure if this is just the tip of the iceberg ...

                                               Cheers,
                                               Michael
                                               [1] http://www.w3.org/TR/2007/WD-cooluris-20071217/#solutions
                                               ----------------------------------------------------------
                                            Michael Hausenblas, MSc.
                                            Institute of Information Systems & Information Management
                                            JOANNEUM RESEARCH Forschungsgesellschaft mbH
                                                http://www.joanneum.at/iis/
                                           ----------------------------------------------------------

                                               -----Original Message-----
                                               From: Leo Sauermann [mailto:sauermann@dfki.uni-kl.de]
                                               Sent: Thursday, December 20, 2007 3:43 AM
                                               To: Hausenblas, Michael
                                               Cc: semantic-web@w3.org; Leo Sauermann
                                               Subject: Re: Cool URIs, the Semantic Web and Everything
                                                      Hi Michael, RDFa people,
                                                      The question is if httpRange-14 [2] is valid in

                                        the case of

                                        XHTML+RDFa.
                                                      The answer is that httpRange-14 is to

                                        distinguish URIs for

                                        information
                                               resources ("web documents") from real-world

                                        objects (the person

                                               "Alice"). As such, it is a recommendation on URIs.
                                                      RDFa is an encoding of RDF, and typically an

                                        RDFa document has two

                                               relations to URIs:
                                               a) the URI of the RDFa document (=the

                                        information resource where I

                                        can
                                               download the RDFa document)
                                               b) the URIs used as subjects, predicates,

                                        objects inside RDF

                                               statements
                                               written inside RDFa documents
                                                      a) is usually a http-200 uri, and a) is an

                                        information resource (=

                                        a
                                               document).
                                               In the rdf statemetns written inside  A, you

                                        would use both URIs

                                        for
                                               real-world objects and information resources.
                                               example (I don't know  rdfa syntax by heart

                                        now, assume this is

                                        rdfa):
                                                      document at www.example.com/homepage/aboutAlice
                                               <html>
                                               <p
                                        rdf:about="http://www.example.com/identifiers/alice#this"
                                        <http://www.example.com/identifiers/alice#this> >
                                                rdf:type foaf:Person.
                                               </p>
                                               <p
                                        rdf:about="http://www.example.com/moreidentifiersusing303/bob"
                                        <http://www.example.com/moreidentifiersusing303/bob> >
                                               rdf:type foaf:Person
                                               </p>
                                               </html>
                                                      assuming this would be valid RDFa, the URI

                                        .../aboutAlice is a

                                               http-return-200 informaiton resource
                                               .../alice#this is a real-world object as it is

                                        not a document (as I

                                               understand timbl on that)
                                               ...303/bob is not intuitively distinguishable -

                                        if you ignore the

                                               rdf:type relation you don't know what it is. So

                                        for this uri you do

                                        a
                                               HTTP get and the server would return a 303

                                        redirect as described in

                                               "cool uris".
                                               once oyu did the 303, you knowthat ....303/bob

                                        is a real world

                                        object.
                                                      so RDFa and 303'/httprange14 are
                                        recommendations caring about
                                               different
                                               angles, 303 is only concerned about URIs, RDFa

                                        about an RDF

                                               serialization. Technically they don't interfere.
                                                      If I would use RDFa much and would like cool

                                        uris, I would go for

                                               #-uris, they are simple to use and easy to

                                        embed in RDFa.

                                               but as shown above, you can use any URI you

                                        want inside rdfa.

                                                      best
                                               Leo
                                                             Hausenblas, Michael schrieb:

                                                   ===
                                                   Disclaimer: Michael, with his
                                        RDFa-Task-Force-member hat off ;)
                                                   ===
                                                              As I gathered "Cool URIs for the
                                        Semantic Web" is a Working

                                               Draft, now.

                                                   Congrats to Leo and his team, great job!
                                                              The following might sound like a naive

                                        question - and I might

                                                   have missed something :) - but: Is TAG

                                        finding httpRange-14 [2]

                                                   equally valid in the case of XHTML+RDFa?
                                                              I've put together some initial thoughts

                                        at the ESWiki [3]

                                                   - any comments welcome!
                                                              Cheers,
                                                       Michael
                                                              [1]

                                              http://lists.w3.org/Archives/Public/public-swd-wg/2007Dec/0103.html

                                                   [2]
                                        http://www.w3.org/2001/tag/issues.html#httpRange-14
                                                   [3] http://esw.w3.org/topic/RDFa_vs_RDFXML
                                                             ----------------------------------------------------------
                                                    Michael Hausenblas, MSc.
                                                    Institute of Information Systems &

                                        Information Management

                                                    JOANNEUM RESEARCH Forschungsgesellschaft mbH
                                                    Steyrergasse 17, A-8010 Graz, AUSTRIA
                                                               <office>
                                                       phone: +43-316-876-1193 (fax:-1191)
                                                      e-mail: michael.hausenblas@joanneum.at
                                                         web: http://www.joanneum.at/iis/
                                                               <private>
                                                      mobile: +43-660-7621761
                                                         web: http://www.sw-app.org/
                                                  ----------------------------------------------------------














                        --
                        ____________________________________________________
                        DI Leo Sauermann       http://www.dfki.de/~sauermann
                        Deutsches Forschungszentrum fuer Kuenstliche Intelligenz DFKI GmbH
                        Trippstadter Strasse 122
                        P.O. Box 2080           Fon:   +49 631 20575-116
                        D-67663 Kaiserslautern  Fax:   +49 631 20575-102
                        Germany                 Mail:  leo.sauermann@dfki.de

                        Geschaeftsfuehrung:
                        Prof.Dr.Dr.h.c.mult. Wolfgang Wahlster (Vorsitzender)
                        Dr. Walter Olthoff
                        Vorsitzender des Aufsichtsrats:
                        Prof. Dr. h.c. Hans A. Aukes
                        Amtsgericht Kaiserslautern, HRB 2313
                        ____________________________________________________
Received on Tuesday, 15 January 2008 11:54:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 July 2008 07:55:27 GMT