NORA: Scalable OWL reasoner based on NoSQL databases and Apache Spark

Tested with OpenJDK 11 and Gradle 7.5.1.

Cite us:

Benítez-Hidalgo A, Navas-Delgado I, Roldán-García MM. NORA: Scalable OWL reasoner based on NoSQL databases and Apache Spark. Softw Pract Exper. 2023; 53(12): 2377–2392. doi: 10.1002/spe.3258

Building the project

Use gradle to build the project:

$ make build

Getting started

Before running the project, edit the config.properties and set your own values. Make sure to set the NORA_CONFIG environment variable to the path of the config.properties file:

$ export NORA_CONFIG=$(pwd)/config.properties

Load the ontology into the database:

$ java -cp nora.jar loader.Loader examples/ontology.owl examples/individuals/ http://w3id.org/examples/ontology.owl

Then, start the reasoning process by running the following command:

$ java -cp nora.jar reasoner.Reasoner

or by using spark-submit:

$ spark-submit \
    --name nora \
    --class reasoner.Reasoner \
    --master spark://host:port \
    --driver-memory 6G \
    --executor-memory 120G \
    file:///home/user/nora.jar 10

A step-by-step guide is available here.

Troubleshooting

Exception java.lang.OutOfMemoryError: Java heap space when running Apache Spark in client mode

The flag -Xmx specifies the maximum memory allocation pool for a Java Virtual Machine (JVM), while -Xms specifies the initial memory allocation pool. These flags can be set to modify the Java memory settings:

• IntelliJ IDEA: Run/Debug configuration > Set VM options to -Xms10g -Xmx80g
• Terminal: Run Java with -Xms10g -Xmx80g

Tests

Run the following command to run the tests:

$ make test

Embedded Cassandra requires JDK 11 or lower. If using a higher version, downgrade it to JDK 11.

License

This project is licensed under the MIT license - see the LICENSE file for details.