Supported Queries

There are currently two query types supported:

• plain text queries
• SPARQL pattern queries

Plain Text Queries

This can be anything: SPARQL, SQL, a whole book if you need to. The only limitation is that it has to fit in one line per query. Every query can be executed as is.

This can be set using the following:

...
    queryHandler:
      className: "InstancesQueryHandler"

SPARQL Pattern Queries

This only works for SPARQL Queries at the moment. The idea came from the DBpedia SPARQL Benchmark paper from 2011 and 2012.

Instead of SPARQL queries as they are, you can set variables, which will be exchanged with real data. Hence Iguana can create thousands of queries using a SPARQL pattern query.

A pattern query might look like the following:

SELECT * {?s rdf:type %%var0%% ; %%var1%% %%var2%%. %%var2%% ?p ?o}

This query in itself cannot be send to a triple store, however we can exchange the variables using real data. Thus we need a reference endpoint (ideally) containing the same data as the dataset which will be tested.

This query will then be exchanged to

SELECT ?var0 ?var1 ?var2 {?s rdf:type ?var0 ; ?var1 ?var2. ?var2 ?p ?o} LIMIT 2000 

and be queried against the reference endpoint.

For each result (limited to 2000) a query instance will be created.

This will be done for every query in the benchmark queries. All instances of these query patterns will be subsummed as if they were one query in the results.

This can be set using the following:

...
    queryHandler:
      className: "PatternQueryHandler"
      endpoint: "http://your-reference-endpoint/sparql"

or

...
    queryHandler:
      className: "PatternQueryHandler"
      endpoint: "http://your-reference-endpoint/sparql"
      limit: 4000