Skip to content

back to TOC

First steps with kgsteward

  1. Clone kgsteward from GitHub.

This is to access the first-steps data and config files. One create an environment variable ${KGSTEWARD_ROOT_DIR} that points to its root directory.

# change dir to where you like to clone kgsteward
git clone https://github.com/sib-swiss/kgsteward.git kgsteward
export KGSTEWARD_ROOT_DIR=`pwd`/kgsteward

  1. Install a supported triplestore.

    If you are new to RDF/SPARQL, you may opt for GraphDB, because of its rich documentation and convenient user interface.

GraphDB Install (the free version of) GraphDB from [Ontotext website](https://www.ontotext.com/products/graphdb/download/?ref=menu), following the vendor instructions. Launch GraphDB, using the application icon or the command line. By default, the user interface of GraphDB becomes available at http://localhost:7200. Alternatively, you may use Docker ...
Fuseki through brew (OSX) 
brew install fuseki
export FUSEKI_DIR=~/scratch/fuseki # FIXME: update path to where you would like to store the db
mkdir -p $FUSEKI_DIR
( cd $FUSEKI_DIR && fuseki-server --config $FIRST_STEPS_DIR/fuseki.config.ttl > $FUSEKI_DIR/logs.txt )&
By default, the user interface of Fuseki becomes available at http://localhost:3030.
RDF4J through docker 
# brew install --cask docker # work on OSX
docker pull eclipse/rdf4j-workbench:5.1.0 # you may try tag ":latest"

export RDF4J_DIR=$HOME/scratch/rdf4j
mkdir -p $RDF4J_DIR
docker run -d \
    -p 8080:8080 \
    -e JAVA_OPTS="-Xms1g -Xmx12g" \
    -v $RDF4J_DIR:/var/rdf4j \
    -v $RDF4J_DIR/logs:/usr/local/tomcat/logs \
    --memory=13G \
    --cpus=3 \
    eclipse/rdf4j-workbench:5.1.0
The user interface becomes available at [http://localhost:8080/rdf4j-workbench](http://localhost:8080/rdf4j-workbench)
QLever through docker QLever manages its own Docker container internally via the `qlever` CLI. You need Docker (or Podman), the `qlever` Python CLI, and Apache Jena's `riot` for RDF format conversion. 
# Install qlever CLI (requires uv or pip)
uv tool install qlever

# Install Apache Jena (provides riot)
brew install jena          # macOS
# or: apt install jena     # Debian/Ubuntu

# Create a working directory for the qlever index
export QLEVER_DIR=$HOME/scratch/qlever/first_steps
mkdir -p $QLEVER_DIR

# Point kgsteward to the provided Qleverfile
export QLEVER_FILE=$KGSTEWARD_ROOT_DIR/doc/first_steps/Qleverfile
QLever exposes a SPARQL endpoint at http://localhost:7019 (as configured in `Qleverfile`).

  1. Install kgsteward

You can install kgsteward globally, following the instructions. Or alternatively, if you have uv installed, you may define an alias which will work as expected only from directory $KGSTEWARD_ROOT_DIRand below

alias kgsteward="uv run $KGSTEWARD_ROOT_DIR/kgsteward"

  1. Create and populate the repository.

The different stores are accessed through different YAML config file. The description of the dataset to be stored is shared by the different configs.

GraphDB 
export GRAPHDB_USERNAME=admin  # default of GraphDB fresh installation
export GRAPHDB_PASSWORD=root   # default of GraphDB fresh installation
cd $KGSTEWARD_ROOT_DIR/doc/first_steps
kgsteward graphdb.yaml -I # rewrite repository
kgsteward graphdb.yaml -C # populate repository
kgsteward graphdb.yaml -V # validate repository
Fuseki 
cd $KGSTEWARD_ROOT_DIR/doc/first_steps
kgsteward fuseki.yaml -I # rewrite repository
kgsteward fuseki.yaml -C # populate repository
kgsteward fuseki.yaml -V # validate repository
RDF4J 
cd $KGSTEWARD_ROOT_DIR/doc/first_steps
kgsteward rdf4j.yaml -I # rewrite repository
kgsteward rdf4j.yaml -C # populate repository
kgsteward rdf4j.yaml -V # validate repository
QLever 
cd $KGSTEWARD_ROOT_DIR/doc/first_steps
kgsteward qlever.yaml -I # reset staging area, copy Qleverfile to QLEVER_DIR
kgsteward qlever.yaml -C # stage all files, build index, start server
kgsteward qlever.yaml -V # validate repository
> **Note:** QLever uses a static index. `-C` stages all RDF files and builds the index in one pass at the end. SPARQL updates from `update:` sections are applied after the index is built and then persisted with a `rebuild-index` step.

Congratulations: you have populated a repository using kgsteward :-)

  1. Details

The following configuration files have been used:

  • dataset.yaml describes the RDF data content of a repository, independently of a particular store engine. This file was manually created. It illustrates the diversity of supported syntaxes.

  • graphdb.yaml describes how to access GraphDB, and includes a link to dataset.yaml. This file was manually edited.

  • graphdb.config.ttl describes the configuration from a GraphDB repository. This file was obtained from the GraphDB user interface. It can be manually modified to some extent. It permits to re-create the same GraphDB repository in another server instance.

  • fuseki.yaml describes how to access Fuseki, and includes a link to dataset.yaml. This file was manually edited.

  • rdf4j.yaml describes how to access RDF4J, and includes a link to dataset.yaml. This file was manually created.

  • rdf4j.config.ttl describes the configuration of a RDF4J repository. This file was exported from the RDF4J user interface, it can be manually modified to some extent. It permits to re-create the same configuration in another server instance.

  • qlever.yaml describes how to access QLever, and includes a link to dataset.yaml. The qlever server location, repository name, and access token are read from the Qleverfile rather than from this YAML.

  • Qleverfile is the native QLever configuration file (INI format). It specifies the dataset name, server port, access token, and Docker image. kgsteward copies this file to $QLEVER_DIR and patches it with MULTI_INPUT_JSON entries before building the index — the original file is never modified.

The full supported syntax of the above YAML files is documented here

back to TOC