[Corpora-List] Re : Looking for free temporal corpora

[Andre Bittar]

Hi Samir

For actual original text data, the TimeBank 1.2 corpus is taken from
the Wall Street Journal corpus (and perhaps other sources, from
memory). For the French corpus, the texts are all from the freely
available Est Républicain corpus
(http://www.cnrtl.fr/corpus/estrepublicain/). In both cases, the
corpus documentation gives details of the original texts used. Or, you
could always remove the XML markup with a simple script...

Another place of interest is the TimeML corpus page:
http://timeml.org/site/timebank/timebank.html

Regards
André


On 27 December 2011 23:53, Samir Bilal <samirbilal2 at yahoo.fr> wrote:
> Hi André,
> Thank you for the links and the paper. Is it possible to have also the
> corpus without the annotation?
>
> Best regards
> Samir
> ________________________________
> De : Andre Bittar <andre.bittar at linguist.jussieu.fr>
> À : Samir Bilal <samirbilal2 at yahoo.fr>
> Cc : "corpora at uib.no" <corpora at uib.no>
> Envoyé le : Mardi 27 Décembre 2011 22h55
> Objet : Re: [Corpora-List] Looking for free temporal corpora
>
> Hi Samir
> If by "temporal corpora" you mean texts annotated with events and the
> temporal relations that hold between them, here are a couple of places
> to look:
>
> http://timeml.org/site/timebank/timebank.html (for English)
> http://www.linguist.univ-paris-diderot.fr/~abittar/french-timebank/ (for
> French)
>
> Also, seeing as you are interested in evaluation, if you haven't
> already done so, I suggest you check out TempEval-2:
> http://www.timeml.org/tempeval2/
> ...and this recent paper:
> http://www.cs.rochester.edu/~naushad/paper/uzzaman_allen11_Temporal_Evaluation.pdf
>
> Cheers
> André
>
>

_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora