CODEWITHSUNDEEP
nlpstanford-nlp
com
com

Reputation: 2704

CoreNLP for Original Dependencies with Neural Network Dependency Parsing

I use Stanford CoreNLP v 3.5.2 in order to get Neural Network Dependency Parsing.

The problem is I don't need Universal Dependencies I need original Dependencies.

Properties file can have parse.originalDependencies attribute, however there two questions

NN DepParsing is depparse option, and the attributes of depparse are not described in http://nlp.stanford.edu/software/corenlp.shtml

I am not sure that depparse.originalDependencies exists.

In addition, how exactly should I define it in properties file

depparse.originalDependencies = 1
or 
depparse.originalDependencies = true

Thank you

Upvotes: 0

Views: 395

Answers (1)

Jon Gauthier
Jon Gauthier

Reputation: 25572

You need to specify a different parsing model using the -depparse.model option in your CoreNLP invocation. (Note that if you are accessing the parser directly through the edu.stanford.nlp.parser.nndep.DependencyParser class, this option is just called -model.)

Use the following -depparse.model option. This refers to a non-Universal Dependencies model that is distributed with the parser.

-depparse.model "edu/stanford/nlp/models/parser/nndep/english_SD.gz"

Upvotes: 1

Related Questions