query over multiple rdf files

Question

Im working on writing sparql queries in my java project using jena API.

I have multiple rdf files, and I need to write query which involves displaying values from multiple rdf files for an attribute.

For a single rdf file, I can load in into a model, create a query string and execute it.

But how do I do it for multiple rdf documents.

Note: All rdf files are locally stored.

My code for single rdf file:

in = new FileInputStream(new File("data.rdf"));
model = ModelFactory.createOntologyModel();
model.read(in,defaultNameSpace);
in.close();

String queryString ="<My SParql Query>";

Query query = QueryFactory.create(queryString);

// Execute the query and obtain results
QueryExecution qe = QueryExecutionFactory.create(query, model);
ResultSet response = qe.execSelect();

// Output query results
ResultSetFormatter.out(System.out, response, query);

// Important - free up resources used running the query
qe.close();

How do I work with multiple rdf files, Can I load them into different models and still use in the same query?? Or should I join the models to get a new model which is union of them?

It's not clear exactly what you're asking. You said "Can I load them into different models and still use in the same query?". Sure, just create a different model, read in a different file, and then the rest of the code is the same. If you want one model that contains the triples from all the files, you could create one model and then use model.read to read in each of the files to the model. Alternatively, as you say, you could create a union model. — Joshua Taylor

Rob Hall Rob Hall · Accepted Answer · 2014-04-16T13:46:14

The answer depends on the type of data that you have. RDF documents represent the logical union of all RDF statements/triples in each of those documents. Taking the union of them is only meaningful if the union of the documents is logically consistent.

Say for example, I have two documents, A and B, whose content are as follows:

:a rdf:type foaf:Person .

.. for A, and and then..

:a rdf:type :Cat .

.. for B. This implies that the union would look like:

:a rdf:type foaf:Person .
:a rdf:type :Cat .

Let us say that your query intends to retrieve the rdf:type for :a, and consider each scenario.

If you were to query the two documents separately, then you would have two different ResultSets, each with one QuerySolution. These two solutions may be logically consistent with your domain, because each document (in your application) should be taken as a completely separate instance of the world.

If you were to query the union, then you would have a single ResultSet with two QuerySolutions, which imply that :a is both a foaf:Person and a :Cat. This may be a violation of your domains logical constraints if, for example, foaf:Person owl:disjointWith :Cat.

Therefore, the question comes back around to your data: Can your data be queried as a single union, or do you need to query it as separate models?

As Joshua Taylor mentioned in his comment, nothing prevents you from crafting a for loop that populates a Model for each document that you wish to query. You can then reuse the same Query object in querying each document. You do, however, need a separate QueryExecution instance for each. Similarly, you could create a Model, then loop over all of your document names and read their content into the same instance (thus, creating the union). Finally, you would execute your query over the unified model.

query over multiple rdf files

1 Answers