MusicBrainz-TERMAL/XML

MusicBrainz rdb2rdf mappings for OBDA Semantika system. Currently we are presenting the native mapping language for Semantika, called TERMAL/XML. In the future the system will be able to read R2RML syntax.

The purpose of this project is to show Semantika's capability in exporting data from database tuples to RDF triples; as well as to show the possible option to reuse R2RML mappings for Semantika. I have included a sample dataset that you can use immediately. No import data required.

Credits for Barry Norton who has provided the R2RML mappings in his repository site LinkedBrainz / MusicBrainz-R2RML. This work is just a pure translation from those mappings.

New Tutorial: How to create a simple SPARQL endpoint for MusicBrainz database.

Preparation:

• To run the examples, you will need to use Semantika CLI project and download the latest release.
• Download it and extract the ZIP file into your local directory.
• Download the mapping files from termalxml/ and place them together with the CLI extraction. Notice that you might have downloaded the configuration file as well (mbzdb.cfg.xml).
• Download the sample dataset from sample-data/ and extract the ZIP into your local directory. The ZIP file contains a self-contained H2 database.
• Browse the H2 home folder and copy-and-paste a JAR file called h2-1.3.xxx.jar to jdbc/ in CLI home folder.

Start Exporting:

• Run the H2 server by executing h2.sh (or h2.bat if you're in Windows) from its home directory.

$ cd $H2_HOME
$ ./h2.sh 

• Run Semantika CLI tool from its home directory.

$ cd $CLI_HOME
$ ./semantika materialize --config=mbzdb.cfg.xml --output=output.n3 -f N3

(Note the other options for the output format, e.g., Turtle, RDF/XML and JSON-LD, see semantika --help)

Important Notes

• All the mappings have been tested using the provided dataset and all passed the run. The total number of triples produced is 1,045,226 Triples.

• Notice that some mappings are enclosed by comment signs. I have put a comment to explain the reason.

• The dataset is a subset of MusicBrainz Database per March 1, 2014. The dataset contains all US artists/groups from 1970 to 1989 who are still active until now. In summary, the dataset consists of 54 tables with 481,401 tuples (~52 MB). The extraction model is defined as below:

artist 
where begin_date_year < 1990 and begin_date_year >= 1970 
  and area = 222 
  and (type = 1 or type = 2)
  and ended = false 

• The dataset is released following the parent license which is under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0.