ml-lab / tupa

Transition-based UCCA Parser

Transition-based UCCA Parser

TUPA is a transition-based parser for Universal Conceptual Cognitive Annotation (UCCA).

Requirements

• Python 3.x
• DyNet

Build

Install the required modules:

git submodule update --init --recursive
virtualenv --python=/usr/bin/python3 .
. bin/activate  # on bash
source bin/activate.csh  # on csh
pip install -r requirements.txt
python -m spacy.en.download all
ci/install-dynet.sh
python ucca/setup.py install
python setup.py install

Train the parser

Having a directory with UCCA passage files (for example, the Wiki corpus), run:

python tupa/parse.py -t <train_dir> -d <dev_dir> -m <model_filename>

To specify a model type (sparse, mlp or bilstm), add -c <model_type>.

Parse a text file

Run the parser on a text file (here named example.txt) using a trained model:

python tupa/parse.py example.txt -m <model_filename>

A file named example.xml will be created.

If you specified a model type using -c when training the model, be sure to include it when parsing too.

Pre-trained models

To download and extract the pre-trained models, run:

wget http://www.cs.huji.ac.il/~danielh/ucca/{sparse,mlp,bilstm}.tgz
tar xvzf sparse.tgz
tar xvzf mlp.tgz
tar xvzf bilstm.tgz

Run the parser using any of them:

python tupa/parse.py example.txt -c sparse -m models/ucca-sparse
python tupa/parse.py example.txt -c mlp -m models/ucca-mlp
python tupa/parse.py example.txt -c bilstm -m models/ucca-bilstm

Author

• Daniel Hershcovich: [email protected]

Citation

If you make use of this software, please cite the following paper:

@inproceedings{hershcovich2017a,
  title={A Transition-Based Directed Acyclic Graph Parser for {UCCA}},
  author={Hershcovich, Daniel and Abend, Omri and Rappoport, Ari},
  booktitle={Proc. of ACL},
  year={2017}
}

License

This package is licensed under the GPLv3 or later license (see LICENSE.txt).