On 04/04/2012 02:37 PM, Jim - FooBar(); wrote:
Then maybe you want something which merges the results of multiple
TokenNameFinders?
Exactly!!! That is what i'm trying to explain and accomplish. However
there is a major issue if the maxent model was trained using the
"drug" (or any other) tag instead of the "default" tag simply because
the dictionary can only be evaluated on a test-set marking the
entities with the "default tag"...
- We could make a generic wrapper around a TokenNameFinder which uses a
Map to rename the produced tags.
- We could work a bit on the DictionaryNameFinder to output a correct tag.
- The class to merge the results would be nice to have in OpenNLP.
Would be nice if you could contribute these things back.
Jörn
