AW: Annotation body a transcription of the target

Dear all,

thanks for the comments on this, which were all very helpful! I'm going with the suggestion below, although I'm leaning towards using rdfs:label instead of cnt:chars, for the reason that Antoine pointed out. P.S.: readers of our blog (http://pelagios-project.blogspot.co.uk) will be able to get their hands on a first live sample of our updated RDF data *really* soon ;-)

Cheers,
Rainer



Von: Robert Sanderson [mailto:azaroth42@gmail.com]
Gesendet: Mittwoch, 16. Oktober 2013 17:26
An: Simon Rainer
Cc: Antoine Isaac; public-openannotation@w3.org
Betreff: Re: Annotation body a transcription of the target


Dear all,

I would think that more annotation clients would do the right thing with the two body method, one toponym with the content in cnt:chars, and one semantic tag with the URI.

So:

<anno1> a oa:Annotation ;
  oa:hasBody _:body1 ;
  oa:hasBody <place1> ;
  oa:hasTarget <target1> ;
  oa:motivatedBy pelagios:someMotivationHere .

_:body1 a cnt:ContentAsText, pelagios:Toponym ;
  cnt:chars "Placename" .

<place1> a oa:SemanticTag, pelagios:PlaceOrSimilarTypeHere .


Using Shared Canvas, the target would be a Canvas representing the physical object and the motivation would be sc:painting, but otherwise it would be identical.

Rob


On Fri, Oct 11, 2013 at 2:51 AM, Simon Rainer <Rainer.Simon@ait.ac.at<mailto:Rainer.Simon@ait.ac.at>> wrote:
Hi Antoine,

yes I agree - it's the URI body that's the semantic tag rather than the description. Otherwise I think I'm seeing a trend towards this solution:

oa:hasBody [ rdf:type pelagios:Toponym ; rdfs:label "Placename"  ]  .
A general question concerning this: is "rdfs:label" preferable over "cnt:chars" then? (Which is what the spec uses for textual bodies.)

Cheers,
Rainer



-----Ursprüngliche Nachricht-----
Von: Antoine Isaac [mailto:aisaac@few.vu.nl<mailto:aisaac@few.vu.nl>]
Gesendet: Freitag, 11. Oktober 2013 08:56
An: public-openannotation@w3.org<mailto:public-openannotation@w3.org>
Betreff: Re: Annotation body a transcription of the target

Hi Rainer,

I'd think the semantic tag is rather the other body, which has been left out of discussion  so far, no?
We have something like

oa:hasBody [ pelagios:theURIForThePlace rdf:type oa:SemanticTag ]  .
oa:hasBody [ rdf:type pelagios:Toponym ; rdfs:label "Placename"  ]  .

The question then is if you want to capture explicitly that the transcription is precisely a transcription of the place denoted by the URI. It doesn't look crucial (unless a same annotation will handle different places at once, which doesn't seem a good thing to do) but *if* you need it, then you'd have two basic choices

a. Representing a direct link between the two bodies:
oa:hasBody [ pelagios:theURIForThePlace rdf:type oa:SemanticTag ]  .
oa:hasBody [ rdf:type pelagios:Toponym ; rdfs:label "Placename" ; pelagios:transcriptionOf pelagios:theURIForThePlace ]  .


b. Mint a specific type for the annotation, to reflect that it's a "semantic annotation con transcription". Which can be done either by subclassing oa:Annotation or introducing a new instance of skos:Concept.

oa:hasBody [ pelagios:theURIForThePlace rdf:type oa:SemanticTag ]  .
oa:hasBody [ rdf:type pelagios:Toponym ; rdfs:label "Placename" ; pelagios:transcriptionOf pelagios:theURIForThePlace ]  .
oa:motivatedBy pelagios:transcribingAndLinkingToGazetteer"

the latter looks a bit uglier perhaps. (and either a and b would be of course more complex than the first solution)

Best,

Antoine


> So then why don't you just use:
>
> hasBody [ chars: "Placename"; rdf:type: "pelagios:Toponym" ] .
>
> Sounds sufficient to me. The type "pelagios:Toponym" seems to imply that it is a transcribed place, right?
>
> Ok, actually now, thinking about it, I understand the problem better.
>
> What about (switching to turtle syntax):
> oa:hasBody [ rdf:type oa:SemanticTag ; rdf:type pelagios:Toponym ; rdfs:label "Placename"  ]  .
>
> Sounds more like a semantic tag to me now.
> Sebastian
>
>
> Am 11.10.2013 07:32, schrieb Simon Rainer:
>> Hi Sebastian,
>>
>> yes, I'd say my options are either 1 or 2. We simply use "pelagios:Toponym" to denote a transcribed place, so option 3 is redundant. (That probably wasn't clear from my last E-Mail...) And Option 4 doesn't happen, since if we have no transcription, we just omit the textual body altogether.
>>
>> Cheers,
>> Rainer
>>
>>
>> -----Ursprüngliche Nachricht-----
>> Von: Sebastian Hellmann [mailto:hellmann@informatik.uni-leipzig.de<mailto:hellmann@informatik.uni-leipzig.de>]
>> Gesendet: Freitag, 11. Oktober 2013 07:10
>> An: Simon Rainer; public-openannotation
>> Betreff: Re: Annotation body a transcription of the target
>>
>> Dear Simon,
>> a clarification question. So your options are:
>>
>> 1.
>> hasBody [ chars: "Placename"; rdf:type: "pelagios:Toponym" ] .
>>
>> 2.
>> hasTranscription [ chars: "Placename"; rdf:type: "pelagios:Toponym" ] .
>> hasTranscription rdfs:subPropertyOf hasBody .
>>
>> 3.
>> hasBody [ chars: "Placename"; rdf:type: "pelagios:Toponym" ; rdf:type "TranscribedPlace" ] .
>>
>> 4. (no transcription)
>> hasBody [ rdf:type: "pelagios:Toponym" ] .
>>
>>
>> Structure-wise these are very similar and I can see no advantage or disadvantage. I think it is a matter of convention.
>> Is there a best practice?
>>
>> All the best,
>> Sebastian
>>
>> Am 10.10.2013 21:36, schrieb Simon Rainer:
>>> Hi,
>>>
>>> as some of you know, the Pelagios project is concerned with annotating place references in different types of documents. Our normal case is that we have one annotation body that is simply a URI representing the place.
>>>
>>> In some cases, however, we also want to attach an actual transcription of the place name as found in the document. To keep annotations coherent (cases without transcription vs. cases with transcription) I'd like to add the transcription as a separate, second body (which should be fine, I guess?).
>>>
>>> Now a quick question/sanity check for the list: I want to explicitely indicate that the textual body is a transcription of a placename. Is the best way to do this to type the body? (The spec only speaks of using typing in terms of media types.) I.e. something like:
>>>
>>> hasBody: [ chars: "Placename"; rdf:type: "pelagios:Toponym" ]
>>>
>>> or should we having our own sub-property of hasBody instead?
>>>
>>> Cheers,
>>> Rainer
>>>
>>>
>>>
>>
>> --
>> Dipl. Inf. Sebastian Hellmann
>> Department of Computer Science, University of Leipzig
>> Events:
>> * NLP & DBpedia 2013 (http://nlp-dbpedia2013.blogs.aksw.org)
>> Venha para a Alemanha como PhD: http://bis.informatik.uni-leipzig.de/csf
>> Projects: http://nlp2rdf.org , http://linguistics.okfn.org , http://dbpedia.org/Wiktionary , http://dbpedia.org
>> Homepage: http://bis.informatik.uni-leipzig.de/SebastianHellmann
>> Research Group: http://aksw.org
>>
>>
>
>