RAKI Verbalizer Pipeline

Install

A conda installation is required.

Install:

conda create --quiet --yes --name rakiEnv python=3.9 && \
conda run --name rakiEnv pip install sentencepiece==0.1.96 subword-nmt==0.3.7 OpenNMT-py==2.3.0 && \
conda activate rakiEnv && conda install  sentencepiece

Download:

cd model && \
wget https://hobbitdata.informatik.uni-leipzig.de/RAKI/VerbalizerModel/model.zip --no-check-certificate  && \
  unzip model.zip && \
  rm model.zip

Compile:

mvn -T 1C compile  -X -am

Run:

ontology="ontology.owl"
axioms="axioms.owl"
out="out.json"
type="model"


export MAVEN_OPTS="-Xmx4G"

eval "$(command conda 'shell.bash' 'hook' 2> /dev/null)"
conda activate rakiEnv

mainClass="org.dice_research.raki.verbalizer.pipeline.ui.RAKICommandLineInterface"

mvn exec:java \
-Dexec.mainClass=$mainClass \
-Dexec.args="-o $ontology -a $axioms -s $out -t $type