On 02/14/2013 05:24 PM, Pankaj Takawale wrote:
Hi All,
I've written simple prototypes for name finding using NameFinderME api, and
for classification using DocumentCategorizerME api.
For both cases, I would like to see how individual feature is impacting the
name finding and classification result.
Also I would like to know weighting being assigned to each feature while
training model.
Is it possible to print this information using OpenNLP api?
As far as I know it is not possible, because there are methods missing
to access this information,
but the information is there.
A contribution to fix this would be very welcome, we could make a util
which prints a model as a csv file
to make it easy to analyze the features with Excel/OpenOffice Spreadsheet.
Jörn
