[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [office-metadata] Question about the IRI of a RDF subject from an ODFdocument
Svante.Schubert@Sun.COM wrote on 02/06/2007 07:57:32 AM: > Hello Elias, > > I think I can speak for all that it is good to see you back on the list. > Your technical experience is as welcome as needed as you just proved > again. I am honestly happy to have you back in the SC. > The usage of xml:base seems to be a very good idea. Thank you. > > I am not sure, if this SC has to solve the following xml:base related > problem, at least it plays into the area of #10 on our list "How do we > cope with changes in content". wow. That item definitely needs more thinking. > > I assume the base URL describes as well the document instance, which is > usually described by no reference at all. > Which means, if the RDF reference of the document instance is being > changed, all relative references to resources in the document are being > changed as well and by this their RDF references. > Meaning that by changing the RDF subject of the document instance, all > RDF subjects of the contained resources are being changed and are > something completely new to RDF applications. > > Let me give an example: > We use an ODF invoice template containing a certain imagine, which stays > the same in all invoices. > I assume that the base URL for the invoice document would change for > every new invoice written. > How would a RDF application is able to figure out that the images in the > document are all the same, when the base URL and by this the image RDF > reference is always different? > > Some might say a possible solution could be to change the base URL with > every save of the document, by this we would be sure to talk about a new > instance. We could use an owl:sameAs to the first URI used by the > template or the URI of the first occurrence of the resource in the document. > On the other hand in the end the user has to define, when the resource > is being changed in a way that it is no longer correct to give an > owl:sameAs. Right. It's definitely complicated. > > My sincere condolences for your loss, Elias. Thank you. > > Regards, > Svante > > Elias Torres wrote: > > Hello everyone, > > > > I'm slowly starting to get back to work, after a long hiatus, but not only > > because of my vacation but due to a death in the family and an additional > > personal accident I had. I hope things have progressed while I was gone and > > that there's still ways in which I can offer some help to the group. Let me > > try for example to suggest some things on the subject of this email. Please > > forgive me if I repeat some suggestion stated before and whether it was > > already made clear that what I'm suggesting either doesn't apply or the > > group decided it was not the right solution. I acknowledge I am not fully > > up to date with all of the discussions that have been going on during my > > absence. > > > > -Elias > > > > Svante.Schubert@Sun.COM wrote on 02/05/2007 03:10:55 PM: > > > > > >> Bruce D'Arcus wrote: > >> > >>> On Jan 30, 2007, at 1:51 PM, Bruce D'Arcus wrote: > >>> > >>> > >>>> This is a trick issue, BTW. Ideally you want a globally unique IRI > >>>> > > > > Definitely a tricky situation. But one that needs to be addressed. First, I > > don't think that there is a simple solution (e.g. an almighty IRI that will > > solve all of the problems). In my opinion, it's just a matter of providing > > enough metadata to describe whatever it is that we are trying to > > communicate/document. > > > > > >>>> with which you can also locate the document. But that's not realistic > >>>> in many desktop scenarios, so I think you're left with two choices: > >>>> > >>>> 1) the one above > >>>> 2) using a URI-encoded UUID > >>>> > >>> Well, and of course using the Flickr URI, which would be good for this > >>> particular case. > >>> > >> By this we have at least one case where the RDF subject IRI is not the > >> reference to the local instance. > >> As we use the flickr URL as the RDF subject IRI, we are in need of a way > >> to reference to the image from our RDF syntax. > >> Only by this, the RDF application would know, that we are talking about > >> a picture in this document. > >> BTW a similar reference would be needed for linking to other ODF > >> elements, for instance if we would like to reference to a certain ODF > >> table or the linked image, that is being used in the document. > >> > >> My question to you now, is it from the RDF point of view sufficient to > >> say, we have this picture somewhere in this document? > >> > >> In N3 something like: > >> <> has <http://www.flickr.com/photos/images/holidayimage.png> . > >> > >> Or are we in need of a package URL to the described content? > >> > >> In N3 something like: > >> @prefix odf: <http://www.oasis-open.org/package-rdf#> > >> <http://www.flickr.com/photos/images/holidayimage.png> odf:pack-ref > >> <odf:/images/holidayimage.png> > >> > >> (A package URL as relative URI references are not allowed in RDF URI > >> References [1] ) > >> > > > > You can't use a relative URI in a triple at the model level, but you can > > use a relative URI in the serialization of triples. > > > > In a document located at /home/eliast/package.n3 you can have the > > following: > > > > <eliast.jpg> dc:title "Elias' mugshot" . > > > > When the document is loaded eliast.jpg would be resolved to the URI: > > "file:///home/eliast/eliast.jpg". > > > > However, we need more than that, because our package doesn't have a stable > > location. Hence, I would suggest we ground it on some URN, like LSID for > > example or whatever. > > > > urn:lsid:www.oasis-open.org:SOME_PACKAGE:SOME_UUID/eliast.jpg > > > > Now if we use RDF/XML we can make use of > > xml:base="urn:lsid:www.oasis-open.org:SOME_PACKAGE:SOME_UUID/". > > > > <eliast.jpg> dc:title "Elias' mugshot" . > > > > For the case that the same exact picture lived on Flickr, then we can use > > owl:sameAs: > > > > <eliast.jpg> owl:sameAs > > <http://www.flickr.com/photos/images/holidayimage.png> . > > > > N3 doesn't have an @base property like SPARQL or RDF/XML, but Turtle which > > is updated by the SWIG, is in the process of adding such extension. > > > > > >>> Might be nice if we could add the ability to have both a path and a > >>> global IRI in the manifest entry for a file. > >>> > >>> > >> [1] Note: Relative URI references are not allowed in RDF URI References: > >> "RDF URI references are compatible with the anyURI > >> <http://www.w3.org/TR/2001/REC-xmlschema-2-20010502/#anyURI> datatype as > >> defined by XML schema datatypes [XML-SCHEMA2 > >> <http://www.w3.org/TR/rdf-concepts/#ref-xml-schema2>], constrained to be > >> an absolute rather than a relative URI reference.", taken from > >> http://www.w3.org/TR/rdf-concepts/#section-Graph-URIref > >> > >> > > > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]