UCCA is a novel linguistic framework for semantic annotation, whose details are available at the following paper:

Universal Conceptual Cognitive Annotation (UCCA)
Omri Abend and Ari Rappoport, ACL 2013

This Python3-only package provides an API to the UCCA annotation and tools to manipulate and process it. Its main features are conversion between different representations of UCCA annotations, and rich objects for all of the linguistic relations which appear in the theoretical framework (see core, layer0, layer1 and convert modules under the ucca package).

Install NLTK and the required modules:

pip3 install --user nltk
python3 -m nltk.downloader averaged_perceptron_tagger punkt

Add this directory to your PYTHONPATH:

setenv PYTHONPATH $PWD

Download and extract the pre-trained model:

wget http://www.cs.huji.ac.il/~danielh/ucca/model.tar.gz
tar xvzf model.tar.gz

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

python3 parsing/parse.py example.txt -m ucca-wiki -s

A file named example.xml will be created.

make dev-install  # creates soft links to the current files
make full-install  # copies the package to the user's python search path

run make help for details

See ucca/README.md for a list of modules under the ucca package.

The scripts package contains various utilities for processing passage files.

The parsing package contains code for a full UCCA parser, currently under construction.

• Amit Beka: amit.beka@gmail.com
• Daniel Hershcovich: danielh@cs.huji.ac.il

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