HeFQUIN FedBench

This project is intended to provide an easy-to-use benchmark for analyzing the behavior of the HeFQUIN engine under various conditions.

Getting started

Download and clean up the FedBench data using:

./prepare_fedbench_datasets.sh

This will take around 15 minutes to complete but only needs to be executed once.

Next, prepared the HDT files that will be used by the LDF servers. Navigate to hdt-converter and run:

docker compose up

This will convert each of the datasets into a single .hdt that can be used by the LDF endpoints.

Running the servers

The FedBench datasets can be hosted using multiple configurations depending on the requirements. The different options are described below:

Run as a single Virtuoso endpoint

The dataset can be run in a single SPARQL endpoint with the datasets made available as named graphs. Navigate to

experiments/virtuoso-named-graphs and run:

docker compose up

This will expose a single Virtuoso endpoint available at http://localhost:8890/sparql. The named graphs are listed below:

• http://example.org/graph/chebi
• http://example.org/graph/dbpedia
• http://example.org/graph/drugbank
• http://example.org/graph/geonames
• http://example.org/graph/jamendo
• http://example.org/graph/kegg
• http://example.org/graph/lmdb
• http://example.org/graph/nyt
• http://example.org/graph/sp2b
• http://example.org/graph/swdfood

This configuration is useful for establishing a source assignment for the FedBench queries, since the named graphs can be bound to variables.

Run as SPARQL endpoints

The datasets can be hosted in separate SPARQL endpoints. Navigate to experiments/virtuoso-endpoints and run:

docker compose up

A proxy container will make the containers available at the following endpoints:

• http://sparql.chebi.localhost
• http://sparql.dbpedia.localhost
• http://sparql.drugbank.localhost
• http://sparql.geonames.localhost
• http://sparql.jamendo.localhost
• http://sparql.kegg.localhost
• http://sparql.lmdb.localhost
• http://sparql.nyt.localhost
• http://sparql.sp2b.localhost
• http://sparql.swdfood.localhost

Run as LDF server endpoints

The datasets can be hosted in separate LDF endpoints that support both TPF and brTPF by default. Navigate to experiments/ldf-endpoints and run:

docker compose up

A proxy container will make the containers available at the following endpoints:

• http://ldf.chebi.localhost
• http://ldf.dbpedia.localhost
• http://ldf.drugbank.localhost
• http://ldf.geonames.localhost
• http://ldf.jamendo.localhost
• http://ldf.kegg.localhost
• http://ldf.lmdb.localhost
• http://ldf.nyt.localhost
• http://ldf.sp2b.localhost
• http://ldf.swdfood.localhost

Run both SPARQL and LDF endpoints

The datasets can be hosted as both SPARQL and LDF endpoints in parallell. Navigate to experiments/all-endpoints and run:

docker compose up

A proxy container will make the SPARQL endpoints available at:

• http://sparql.chebi.localhost
• http://sparql.dbpedia.localhost
• http://sparql.drugbank.localhost
• http://sparql.geonames.localhost
• http://sparql.jamendo.localhost
• http://sparql.kegg.localhost
• http://sparql.lmdb.localhost
• http://sparql.nyt.localhost
• http://sparql.sp2b.localhost
• http://sparql.swdfood.localhost

A proxy container will make the LDF endpoints available at:

• http://ldf.chebi.localhost
• http://ldf.dbpedia.localhost
• http://ldf.drugbank.localhost
• http://ldf.geonames.localhost
• http://ldf.jamendo.localhost
• http://ldf.kegg.localhost
• http://ldf.lmdb.localhost
• http://ldf.nyt.localhost
• http://ldf.sp2b.localhost
• http://ldf.swdfood.localhost