Shape Index Solidbench Experiment with Comunica where 80% of the datasets have a shape index

Experiments to benchmark solidbench queries using the shape index approach, where 80% of the datasets expose a shape index. The queries are available in the ./queries directory. This experiment tests a setup where each dataset is described maximally using shapes and a shape index.

Dependencies

• Nodejs version 22

Has only been tested on Linux

Installation

Make sure that all the submodules are correctly installed

Running the experiment

First, the solidbench dataset must be generated and served locally.

pushd ./server
    ./install.sh
    yarn run solidbench-generate
    yarn run solidbench-serve
popd

Second, the experiment must be run through the benchmark runner. To do so, first install comunica

pushd ./simple-comunica-runner
    ./install.sh
popd

second run, the benchmark runner

pushd ./simple-solidbench-comunica-runner
    yarn install
    yarn node index.mjs -q ../queries -c ../config.json -r 50 -e ../simple-comunica-runner/index.mjs -o ../results -n "standard-shape-index-experiment" &> ../results/log
popd

you can save each iteration of the benchmark to a nextcloud cloud storage using the -b parameter with the URL of a cloud storage with a write permission. The results will be materialized in the ./results directory.