evelinacs / semantic_parsing_with_IRTGs

Experiments of developing an IRTG which simultaneously encodes transformations between phrase structure trees, dependency graphs and semantic graphs.

semantic_parsing_with_IRTGs

This repository contains our experiments of developing an IRTG (Interpreted Regular Tree Grammar) which implements a mapping between the output of the Stanford Parser, Universal Dependencies v2.1 and 4lang.

Our system for Surface Realization Shared Task 2019 can be found here.

Python virtual environment

Whenever possible, all code should be written in Python3 and use the repository's virtual environment.

Creating the virtual environment

With virtualenvwrapper

mkvirtualenv -p python3 irtg

Without virtualenvwrapper

python3 -m venv .venv

If the default python executable is Python3 (e.g. on Windows):

python -m venv .venv

Activating the virtual environment

With virtualenvwrapper

workon irtg

Without virtualenvwrapper

Linux

. .venv/bin/activate

Windows

Using CMD.exe:

.venv\Scripts\activate.bat

Using powershell:

.\.venv\Scripts\Activate.ps1

Installing the required modules

After activating the virtual environment:

pip install -r requirements.txt

Deactivating the virtual environment

deactivate

Alto console usage

java -cp "<path to ALTO's jar>" de.up.ling.irtg.script.ParsingEvaluator "<path to input file>" -g "<path to the grammar file>" -I "<input format>" -O "<output format>" -o "<output file>"