W3C home > Mailing lists > Public > public-lod@w3.org > February 2013

Re: annotations and RDF

From: Matteo Casu <mattecasu@gmail.com>
Date: Thu, 7 Feb 2013 15:39:04 +0100
Cc: Robert Sanderson <azaroth42@gmail.com>, public-lod@w3.org, semantic-web@w3.org
Message-Id: <688B9B6B-1841-413E-BDE0-8561625684E5@gmail.com>
To: Paolo Ciccarese <paolo.ciccarese@gmail.com>
Hi Paolo,

my concerning was about the "info for detecting the text fragment": do you use a begin/end approach on the document (as in UIMA) without storing the body of the text? If it is so, then everything is clear to me.
Forgive me, my point of view is still unripe on the subject, so probably I'm just getting caught into a false problem.. :-)




Il giorno 07/feb/2013, alle ore 14:54, Paolo Ciccarese <paolo.ciccarese@gmail.com> ha scritto:

> Hi Matteo,
> in the Domeo Annotation Tool http://annotationframework.org we do exactly that. We create annotation on text fragment(s), images, tables and we store the annotation, together with the info for detecting the text fragments in a RDF in a separate store. In fact, most of hte times we do not control the pages we are looking at. We also use CiTO and FaBIO for storing the bibliographic data and those are based on FRBR. 
> 
> Could you give me a concrete example of the duplication problem you are mentioning at the end? 
> 
> Best,
> Paolo
> 
> On Thu, Feb 7, 2013 at 8:43 AM, Matteo Casu <mattecasu@gmail.com> wrote:
> Thank you Robert!
> 
> I've just seen what I think is the new draft (february 5th). I will go through it! In the meantime, I'm wondering what you think on the problem of keeping all the annotations of a text in RDF vs. keeping them in a separate store and bind them to entities in the RDF.
> 
> The use case I have in mind is: imagine a book, say The Lord of the Rings. Assume we want to annotate domain information in RDF (characters, actions, etc..) as well as linguistic (or "librarian")-oriented annotations: paragraphs, lines, pages (in order to make citations..), down to lemmas and so on..
> 
> We could follow the FRBR model and keep in an RDF graph the domain information AND some librarian information. But what about the annotations on text as -- say -- links between a character and the lines on which they appear?Should these be RDF statements? What about the the problem of text duplications in annotations which are not independent (e.g. lemmas and sentences)?
> Have you (as a community) a definite idea on this issue or perhaps is something which is still under observation?
> 
> 
> 
> 
> Il giorno 04/feb/2013, alle ore 21:37, Robert Sanderson <azaroth42@gmail.com> ha scritto:
> 
> > Hi Matteo,
> >
> > The Annotation Ontology has merged with Open Annotation Collaboration
> > in the W3C community group:
> >  http://www.w3.org/community/openannotation/
> >
> > And Paolo is co-chair along with myself.
> >
> > We're *just* about to release the next version of the Community Group
> > draft, so your interest comes at a great time.
> > The NIF folk are also part of the Community Group, and we of course
> > would encourage your participation as well!
> >
> > Many thanks,
> >
> > Rob Sanderson
> > (Open Annotation Community Group co-chair)
> >
> >
> > On Mon, Feb 4, 2013 at 2:55 AM, Matteo Casu <mattecasu@gmail.com> wrote:
> >> Hi everybody,
> >>
> >> [my apologies for cross posting -- possibly of interest for both communities]
> >>
> >> does anybody could point me to the major pros and cons in using the Annotation Ontology [0] [1] vs. the NLP interchange format in the context of annotating (portions of) literary texts? My impression is that when someone is using UIMA, the integration of AO with Clerezza-UIMA could give more comfort wrt NiF.
> >>
> >> [0] http://code.google.com/p/annotation-ontology/
> >> [1] http://www.annotationframework.org/
> >> [2] http://nlp2rdf.org/about
> >>
> 
> 
> 
Received on Thursday, 7 February 2013 14:39:38 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:46 UTC