Change default variant to new "rnn" where possible

dkpro / dkpro-core

Collection of software components for natural language processing (NLP) based on the Apache UIMA framework.

https://dkpro.github.io/dkpro-core

Other

196 stars 67 forks source link

Change default variant to new "rnn" where possible #361

Open reckart opened 9 years ago

reckart commented 9 years ago

The StanfordParser supports a new model based on Compositional Vector Grammars. These
are comparative to the factored parser and faster. Should we switch the default variant
from factored to rnn where rnn models are available? (English / Chinese)

http://nlp.stanford.edu/pubs/SocherBauerManningNg_ACL2013.pdf

Original issue reported on code.google.com by richard.eckart on 2014-03-07 14:31:19

reckart commented 9 years ago

I am for switching to rnn models if available;
the paper says that the performance is better and the parser is 20% faster (English)

Original issue reported on code.google.com by eckle.kohler on 2014-03-07 14:47:50

reckart commented 9 years ago

(No text was entered with this change)

Original issue reported on code.google.com by richard.eckart on 2014-03-26 10:51:39

Labels added: Milestone-1.7.0
Labels removed: Milestone-1.6.0

reckart commented 9 years ago

(No text was entered with this change)

Original issue reported on code.google.com by richard.eckart on 2014-08-14 12:04:04

Labels added: Milestone
Labels removed: Milestone-1.7.0

reckart commented 9 years ago

(No text was entered with this change)

Original issue reported on code.google.com by richard.eckart on 2014-08-14 14:18:41

Labels removed: Milestone