RDF data management and processing

Setting up a triplestore

There are many triplestores. The simplest to set up is probably Fuseki.

• Download the archive for Apache Jena Fuseki from Jena download page.
• In the archive, there is an executable: fuseki-server.bat for Windows systems, fuseki-server for Unix-based systems. Execute it. The server will be running in the background.
• With your Web browser, go to http://localhost:3030. This interface allows you to manage your data.
• Go to "manage datasets". Create a new dataset and make it persistent.
• Upload an RDF file of your choice.

Generate RDF with Apache Jena

These instructions assume that you are programming in Java, preferably with Eclipse, using the Apache Jena libraries. You may also use RDFlib in Python, or Redland RDF libary in C, or dotNetRDF in C♯, or EasyRDF for PHP, or N3.js for JavaScript, or Ruby RDF for Ruby, or SWI-Prolog Semantic Web Library, etc.

These operations should get you started with Apacha Jena and Eclipse. If you are using a different library, look at the documetation.

• Start Eclipse. It is available on the Windows system from the Windows "Start" menu.
• Go to File -> New -> Java Project....
• Use the text box to search for "Maven Project". Select it and click Next >.
• Check the box "Create a simple project (skip archetype selection)". Click Next >.
• In the Artifact's Group Id, write fr.emse.master. In the Artifact's Artifact Id, write semweb. Click Finish.
• Eclipse will generate a project with a folder structure and one file called pom.xml. Double click on this file.
• You should see information about the file and at the bottom, tabs named "Overview", "Dependencies", "Dependency Hierarchy", "Effective POM", and "pom.xml". Click on "pom.xml".
• Replace the existing code there with the one in this pom.xml. If you used a different groupId or artifactId, change it accordingly.

Now you will generate RDF data from non-RDF sources. Use the Jena tutorial to familiarise yourself with the API and learn how to generate an RDF graph programmatically.

• Download the TER dataset (or just download the files from this folder). This dataset is published as open data by SNCF and describes the regional train lines, stations and schedule.
• Look at the file stops.txt in the dataset you downloaded. It just describes the names and location of train stations. What is relevant is the stop_id, stop_name, stop_lat and stop_lon.
• Use Jena to make an RDF graph that describes the names and locations of the train stations from the file mentioned before. Geolocated things can be described as instances (rdf:type) of the class http://www.w3.org/2003/01/geo/wgs84_pos#SpatialThing, usually abbreviated as geo:SpatialThing. The WGS84 Geo Positioning vocabulary also provides RDF properties for latitude (geo:lat) and longitude (geo:long). Generate IRIs for each stops based on their stop_id.
• From a single line of the file, the resulting RDF should be (in Turtle):

  @prefix ex: <http://www.example.com/> .
  @prefix geo: <http://www.w3.org/2003/01/geo/wgs84_pos#> .
  @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
  @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
  
  ex:StopArea:OCE80194035  a  geo:SpatialThing;
  	rdfs:label  "gare de Neustadt (Weinstr) Hbf"@fr;
  	geo:lat  "49.35006155"^^xsd:decimal;
  	geo:long "8.14067588"^^xsd:decimal .

You can generate all the data at once in a large Jena Model and serialise it as RDF, or you can fill in a triplestore little by little. If you want to add data to a triplestore such as Jena Fuseki, you can send update queries like this:

Model model = ModelFactory.createDefaultModel();
// ... build the model
String datasetURL = "http://localhost:3030/dataset";
String sparqlEndpoint = datasetURL + "/sparql";
String sparqlUpdate = datasetURL + "/update";
String graphStore = datasetURL + "/data";
RDFConnection conneg = RDFConnectionFactory.connect(sparqlEndpoint,sparqlUpdate,graphStore);
conneg.load(model); // add the content of model to the triplestore
conneg.update("INSERT DATA { <test> a <TestClass> }"); // add the triple to the triplestore

If you finish fast, you can then try to define a vocabulary for GTFS and transform all the SNCF data to RDF.