I know several pipelines (GATE and several commercial NLP pipelines) annotate dates, measurements, address etc http://gate.ac.uk/sale/tao/splitap6.html#x36-736000F.7.
I would like to have the same type of rule/re pattern annotation capability in a Stanbol chain. I would rather not just throw in GATE, but am thinking that perhaps the regex primitive in the Jena rule engine or using constructive SPARQL statements could achieve the same capability with existing core components. Obviously such an engine could utilize language/locale data to select pattern variants and to some degree the POS annotations. What ideas or feedback do people have on this? for(;;); /* eantel...@gmail.com */
