multi-extract: Scripts for extracting multi-synchronous context free grammars

by Graham Neubig ([email protected])

Usage

This is a package for extracting multi-sychronous context free grammars, described in the paper "Multi-target Machine Translation with Multi-synchronous Context Free Grammars" presented in NAACL 2015. The main scripts are:

• train-multi.pl: Full training scripts
• multi-extract.py: A script for extracting the full grammar

Full scripts to reproduce the experiments in the paper can be found here: http://www.phontron.com/project/naacl2015/

Some of the extraction scripts rely on the Travatar toolkit: http://www.phontron.com/travatar/