Hi!

If you have an UIMA pear file, now you can make it work in Stanbol
trough HTTP Rest. A detailed doc is here:
http://pedia.sztaki.hu/?page_id=154 (temporarily under Sztakipedia
Worpress, but I think we should find a better place for it)

Summary:
-I have adapted the CasLight code from Sztakipedia that is a
lightweight implementation for UIMA-like Feature Structures and
packaged it to a bundle.
-The UIMA Remote Client Enhancement Engine uses this code to contact
UIMA SimpleServlets and puts it in a Content Part
-The UIMA To Triples Enhancement Engine gets the stuff from the
dedicated content part and turns certain UIMA Annotations to RDF
triples in a configurable way.

Everything that has been developed is here:
http://pedia2.sztaki.hu/stanbol/

These are the direct links for the UIMA Engines I used for testing
(feel free to play with them!)
http://pedia2.sztaki.hu:8080/engpostagger/
http://pedia2.sztaki.hu:8080/snowball/
http://pedia2.sztaki.hu:8080/languagerec/

And here is the Stanbol instance that uses those endpoints:
http://pedia2.sztaki.hu:9090/enhancer

The UIMA To Triples filters the results and only processes those which
posTag matches n.* (noun types), or those which lemma is ‘not’.
TokenAnnotations are translated to Noun type TextAnnotations in the
output, posTags are translated to sso:posTags

Local UIMA (stanbol and UIMA in the same JVM) is still a pain because
of classloading problems, if you want to just use a pear file and not
compile a whole bundle on your own. Things look a bit better if you
are prepeared to compile both your uima project and the stanbol bundle
(meaning that you have the UIMA SDK in eclipse+stanbol deps). Will
make further efforts about this.

Please, try it out! If you have any questions, suggestions let me know!

Mihály

Reply via email to