GRAPHIA QLever Evaluation: benchmark environment

Benchmark environment used for the RDF backend evaluation carried out in the GRAPHIA project, comparing QLever and Virtuoso on a synthetic knowledge graph reproducing the structural profile of the GoTriple knowledge graph.

The generator produces 8,000,000 documents, 1,000,000 datasets, 500,000 media objects, 100,000 semantic artefacts, 500,000 author profiles and 10,000 projects, of which the first 150,000 documents carry citation links. As reported in the accompanying evaluation report, this yields 155,554,861 triples across 10,110,005 named graphs.

Contents:

• src/generate_kg.py - synthetic knowledge graph generator
• scripts/init.sh - benchmark orchestration
• scripts/qlever_persistance_test.sh - update persistence tests
• scripts/reload.sh - index reload helper
• qlever-evaluation.ipynb - evaluation notebook
• docker-compose.yml, data/ - QLever and Virtuoso configurations

This is the software supplement to the QLever Evaluation Report .