Skip to content
ilija139 edited this page Sep 25, 2014 · 1 revision

For training you will need the PDTB corpus and its dependency trees. See the src/variable.rb for all the required files. To train the parser with different features use the train method in src/trainer.rb. Six models are trained: Connective, ArgPos, ArgExtSS, Explicit, Implicit and Attribution, each in corresponding class. To toggle features change the initialise method in these classes.

Clone this wiki locally