Running Blazegraph

In the following we assume that you want to run Blazegraph on the knowledge base Fbeasy. To run on one of the other knowledge bases, replace fbeasy by freebase or wikidata in all the commands below.

• Download the index + executable files (use the following command, alternatively they are linked on the main page). These contain everything that is needed to run Blazegraph on all three knowledge bases.

  wget https://qlever.cs.uni-freibug.de/cikm2022/blazegraph-indexes.tar
  

• Extract the index files

  tar xvf blazegraph-indexes.tar

• Make the folder read- and writeable for everyone (you need at least read and write permissions for the Docker user) and enter it.

  chmod a+rwx blazegraph-indexes
  chmod a+rw blazegraph-indexes/*
  cd blazegraph-indexes
  

• Pull the Openjdk Docker Image and use it to run Blazegraph.

  This assumes that the port 9999 is free on your machine. Otherwise, change the export PORT=9999 command to an available port.

  docker pull openjdk:8
  docker build -t blazegraph .
  export KB=fbeasy; export PORT=9999;
  docker run -it -d --restart=unless-stopped -e KB=$KB-e PORT=$PORT -p $PORT:$PORT -v $(pwd):/data --name blazegraph.$KB blazegraph
  

• To check the startup process of Blazegraph you can use docker logs

  docker logs -f blazegraph.fbeasy
  

• After a few seconds the server should be ready (the log should say so). You can check this by entering http://localhost:9999/bigdata which shows a UI, in the query tab you can then enter the following query:

  SELECT (COUNT(*) AS ?cnt) WHERE { ?s ?p ?o }
  	  

  The Fbeasy instance should report over 300 Million triples. For Freebase and Wikidata the number of triples should be 1.9 Billion (Freebase) and 6.9 Billion (Wikidata).

  If this doesn't work, also try 172.17.0.1 (Docker's virtual network bridge) instead of localhost, depending on your Docker configuration.

• The Blazegraph-Fbeasy instance is now ready. You can run the evaluation script as described in the respective tutorial.