Distributed Computation of RDF Dataset Statistics
For further development of the projet see https://github.com/SANSA-Stack/SANSA-RDF, since it has been integrated with SANSA core.
Over the last years, the Semantic Web has been growing steadily. Today, we count more than 10,000 datasets made online available following Semantic Web standards. Thanks to those standards, data is machine-readable and interoperable. Nevertheless, many applications, such as data integration, search, and interlinking, may not take the full advantage of the data without having a priori statistical information about data's internal structure and coverage. In fact, there are already a number of tools, which offer such statistics, providing basic information about RDF datasets and vocabularies. However, those works showed severe deficiencies in terms of performance once the dataset size grows beyond the capabilities of a single machine. In this paper, we introduce an approach for statistical calculation of large RDF datasets, which scales out to clusters of machines. We describe the first distributed in-memory approach for computing 32 different statistical criteria for RDF datasets using Apache Spark. The preliminary results show that our distributed approach improves upon a previous centralized approach we compare against.
The main application class is net.sansa_stack.rdf.spark.RDFStats.
The application requires as application arguments:
- path to the input folder containing the RDF data as nt (e.g.
/data/input) - path to the output folder to write the resulting to (e.g.
/data/output)
All Spark workers should have access to the /data/input and /data/output directories.
To run the application on a standalone Spark cluster
- Setup a Spark cluster
- Build the application with Maven
cd /path/to/application
mvn clean package
- Submit the application to the Spark cluster
spark-submit \
--class net.sansa_stack.rdf.spark.RDFStats \
--master spark://spark-master:7077 \
/app/application.jar \
-i /data/input -o /data/output
To run the application, execute the following steps:
- Setup a Spark cluster as described on http://github.com/big-data-europe/docker-spark.
- Build the Docker image:
docker build --rm=true -t sansa/spark-rdf-statistics . - Run the Docker container:
docker run --name Spark-RDF-Statistics-app -e ENABLE_INIT_DAEMON=false --link spark-master:spark-master -d sansa/spark-rdf-statistics
Spark/HDFS Workbench Docker Compose file contains HDFS Docker (one namenode and two datanodes), Spark Docker (one master and one worker) and HUE Docker as an HDFS File browser to upload files into HDFS easily. Then, this workbench will play a role as for Spark-RDF-Statistics application to perform computations. Let's get started and deploy our pipeline with Docker Compose. Run the pipeline:
docker network create hadoop
docker-compose up -d
First, let’s throw some data into our HDFS now by using Hue FileBrowser runing in our network. To perform these actions navigate to http://your.docker.host:8088/home. Use “hue” username with any password to login into the FileBrowser (“hue” user is set up as a proxy user for HDFS, see hadoop.env for the configuration parameters). Click on “File Browser” in upper right corner of the screen and use GUI to create /user/root/input and /user/root/output folders and upload the data file into /input folder. Go to http://your.docker.host:50070 and check if the file exists under the path ‘/user/root/input/yourfile.nt’.
After we have all the configuration needed for our example, let’s rebuild Spark-RDF-Statistics.
docker build --rm=true -t sansa/spark-rdf-statistics .
And then just run this image:
docker run --name Spark-RDF-Statistics-app --net hadoop --link spark-master:spark-master \
-e ENABLE_INIT_DAEMON=false \
-d sansa/spark-rdf-statistics
#Evaluations For more details on evaluations and experiments, have a look on the evaluation section.