python-dependency-parser

An example dependency parser for NLP, use arc-eager algorithm.

Requirement

1. libsvm (with python interface)
2. Python NLTK package, now used to draw tree

Usage

Parse

python parser.py --test to check predict accuracy with d.data

python parser.py file to show dependency tree of given sentences from file, format:

word POS-tag
word POS-tag

sentence should be separated by empty line

It may take seconds for script to load model

Train

1. python trainer.py input out to covert input file to features list, save to output file.

2. use svm-train to train features to model

   now suggest params with -c 32 -g 0.5. You may want to try to use lib linear to train features file to model, which is much faster.