On 10/7/11 4:27 PM, Rooney, Niall wrote:
I am running OpenNLP parser over the movie review sentiment data set and I am receiving a
number of "Could'n't find parse for" messages. Is there limitations to what
can be parsed based on the most recent version of the tools jar and
en-parser-chunking.bin model?
I also saw your post over at SourceForge, but we actually moved all
activities
over to Apache.
The parsing algorithm might not always be able to find a parse for a
sentence.
In this case the mentioned error message is printed to the console (yes,
error
handling in this case should be better). There is no limitation in a way
shareware
software does it.
Can you post one of your sentences here?
The parser was trained on news articles, you might want to add training
data based on
movie reviews.
Jörn
