The component that we have now accepts one model at a time. For the OpenNLP
name finder, it does make sense to run multiple models at the same time,
because each model annotates only exactly one kind of named entity. It's
unclear, however, how to package the resources in that case or how to load them.
Looks like there may be an experimental support in OpenNLP for models
containing multiple labels too.
Original issue reported on code.google.com by richard.eckart on 1 Sep 2013 at 9:50
Original issue reported on code.google.com by
richard.eckart
on 1 Sep 2013 at 9:50