2011/7/18 Rupert Westenthaler <[email protected]>: > Hi Florent > > On Mon, Jul 18, 2011 at 11:09 PM, florent andré > <[email protected]> wrote: >> Hi ! >> >> I worked on the UIMA engine to make it more generic. >> It's now easy to add uima annotator, and I try the RegexAnnotator [1]. >> >> Depending on the configured regex, this annotator can output email, isbn,... >> >> AFAIK there is for now just TextAnnotation and EntityAnnotation type, and I >> don't know if they are suitable for things like email, telephone number,... >> > I see several possibilities: > > 1) use a TextAnnotation with a custom value for dc:type > > urn:123 rdf:type TextAnnotation > urn:123 dc:type <http://www.w3.org/2006/vcard/ns#Cell> > urn:123 selected-text "+43 655 290989" > urn:123 start "123"^^xsd:int > urn:123 end "137"^^xsd:int > > I used here the concepts defined for CellPhones by the vCard ontology > > 2) use a TextAnnotation with an additional type > > urn:123 rdf:type TextAnnotation > urn:123 rdf:type http://schema.org/ContactPoint > urn:123 selected-text "+43 655 290989" > urn:123 start "123"^^xsd:int > urn:123 end "137"^^xsd:int > urn:123 http://schema.org/telephone "+43655290989" > > Here I used the ContactPoint as defined by schema.org > > As I am writing this I have a preference for variant (2). Any other > opinions, suggestions?
urn:123 should not have both types at the same time: it's not true that the annotation is the ContactPoint. I would rather use: urn:123 rdf:type TextAnnotation urn:123 selected-text "+43 655 290989" urn:123 start "123"^^xsd:int urn:123 end "137"^^xsd:int urn:123 dc:related urn:456 urn:456 rdf:type http://schema.org/ContactPoint urn:456 http://schema.org/telephone "+43655290989" In the future, we should define a dedicated property to replace dc:related so as to express: urn:456 "is a suggested semantic interpretation of the 'stuff' referenced by the text annotation" urn:123. There might be several possible interpretation with confidence various confidence score in case of ambiguity. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel
