MYahiatene/ontology-lexicalization_de

Bridging the gap between natural language and knowledge base

Automatic induction of lexicalization for multilingual question answering over linked data

Pipeline exists of 2 major steps:

• Creating association rules for specified classes with extended LexExMachina (takes about 1 week for all classes)
• Automatic induction of association rules to csv files for nouns verbs and adjectives (takes about a few hours for all classes)

Prerequisites

• About 7gb of disk space for the docker container

• About 6GB memory resource allocation inside the docker container

• Docker version >= 20

Execute the System (docker)

pull the latest image:

docker pull myahiatene/ontology-lexicalization_de

You can either run the project for all 354 classes or explicitly defining the classes to be run:

start the container for all classes:

docker run --rm -v <path on your local machine where your results should be stored>:/app/post_processing/csv_results myahiatene/ontology-lexicalization_de

Example of 2 classes: docker run --rm -v <path on your local machine where your results should be stored>:/app/post_processing/csv_results myahiatene/ontology-lexicalization_de "Person" "Actor"

Developers

• Mokrane Yahiatene

Supervisors

• Dr. Philipp Cimiano
• Mohammad Fazleh Elahi

---