Skip to content

Latest commit

 

History

History
139 lines (96 loc) · 6.03 KB

README.md

File metadata and controls

139 lines (96 loc) · 6.03 KB

Computing Recursive SPARQL Queries

Introducing Recursion in SPARQL

We developed a SPARQL function called wfn:runSPARQL that takes a string containing a SPARQL query as input and executes it. Such SPARQL query can refer to itself, thus enabling recursion and extending the range of computable functions within SPARQL. It allows simple recursive computations such as factorial or graph traversal algorithms to be easily implemented within SPARQL on Fuseki.

Our work can be found in:

Install and compile (tested with Fuseki 3.8.0)

  1. ensure you have OpenJDK 8 or later correctly installed
  2. download and extract Apache Fuseki 2 on directory ./fuseki
  3. compile runSPARQL Java source code: ./compile
  4. run fuseki using testing settings: ./run-fuseki. The command will use the the following default configurations:
    • fuseki server settings in config/config.ttl
    • initial data in config/dataset.ttl
    • log4j settings in config/log4j.properties
  5. go to: http://127.0.0.1:3030 and run a recursive query (see below for examples)

Usage

Officially, we suggest the use of the following namespace:

PREFIX wfn: <http://webofcode.org/wfn/>

The function is called with wfn:runSPARQL. Please note that if function is not registered, you must use PREFIX wfn: <java:org.webofcode.wfn.>.

The usage is the following:

 wfn:runSPARQL(query, endpoint [,i0, i1, ...])

where:

  • query is a SPARQL query fragment that will be executed by the runSPARQL function
  • endpoint is a url pointing to the SPARQL endpoint that must be used to run the SPARQL query
  • i0, i1, etc are optional parameters to be used in the SPARQL query

When called, the runSPARQL custom function will execute query (that may contain references to runSPARQL itself) against endpoint, where variables ?i0, ?i1, etc are binded to the values passed to runSPARQL.

Note that the query is expected to produce one result only. If there are more results, only the first one provided by the endpoint is returned (blank node are filtered out).

Examples

Computing the Factorial

The following is an example of recursive SPARQL query that computes the factorial of 3, returning "6"^^xsd:integer:

PREFIX wfn: <http://webofcode.org/wfn/> 

SELECT ?result { 
        # bind variables to parameter values 
        VALUES (?query ?endpoint) { ( 
                "BIND ( IF(?i0 <= 0, 1, ?i0 * wfn:runSPARQL(?query,?endpoint, ?i0 -1)) AS ?result)" 
                "http://127.0.0.1:3030/ds/sparql"
        )}
   
        # actual call of the recursive query 
        BIND( wfn:runSPARQL(?query,?endpoint,3) AS ?result)
}

At the first iteration, runSPARQL will run the following:

PREFIX wfn: <java:org.webofcode.wfn.>
SELECT ?result {
    # bind variables to parameter values
    VALUES (?query ?endpoint ?i0) { (
        "BIND ( IF(?i0 <= 0, 1, ?i0 * wfn:runSPARQL(?query,?endpoint, ?i0 -1)) AS ?result)"
        "http://127.0.0.1:3030/ds/sparql"
        3
    )}
    
    # the recursive query
    BIND ( IF(?i0 <= 0, 1, ?i0 * wfn:runSPARQL(?query,?endpoint, ?i0 -1)) AS ?result)
    FILTER (!isBlank(?result))
} LIMIT 1

In the following iteration ?i0 will be then set to 2, ..., until 0 is reached and recursion stops as per the base case specified in ?query.

Graph search

An example of recursive SPARQL query that computes the distance between two nodes (dbo:PopulatedPlace and dbo:Village) in a hierarchy:

PREFIX wfn: <http://webofcode.org/wfn/>

SELECT ?result { 
        # bind variables to parameter values 
        VALUES (?query ?endpoint) { ( 
                "?i0 <http://www.w3.org/2000/01/rdf-schema#subClassOf> ?next. BIND( IF(?next = <http://dbpedia.org/ontology/PopulatedPlace>, 1 , 1 + wfn:runSPARQL(?query, ?endpoint, ?next)) AS ?result)" 
                "http://127.0.0.1:3030/ds/sparql"
        )}
   
        # actual call of the recursive query 
        BIND( wfn:runSPARQL(?query,?endpoint, <http://dbpedia.org/ontology/Village>) AS ?result)
} 

Note that the default dataset is used (in config/dataset.ttl) where dbo:Village and dbo:PopulatedPlace have distance 2 like in DBpedia:

dbr:Village -> dbr:Settlement -> dbr:PopulatedPlace

The call of runSPARQL will return "2"^^xsd:integer, by generating another SPARQL query, recursively calling runSPARQL, that in its first iteration is similar to the following one:

PREFIX wfn: <java:org.webofcode.wfn.>
SELECT ?result {
    # bind variables to parameter values
    VALUES (?query ?endpoint ?i0) { (
        "?i0 <http://www.w3.org/2000/01/rdf-schema#subClassOf> ?next. BIND( IF(?next = <http://dbpedia.org/ontology/PopulatedPlace>, 1 , 1 + wfn:runSPARQL(?query, ?endpoint, ?next)) AS ?result)"
        "http://127.0.0.1:3030/ds/sparql"
        <http://dbpedia.org/ontology/Village>
     )}
    
    # the recursive query
    ?i0 <http://www.w3.org/2000/01/rdf-schema#subClassOf> ?next. 
    BIND( IF(?next = <http://dbpedia.org/ontology/PopulatedPlace>, 1 , 1 + wfn:runSPARQL(?query, ?endpoint, ?next)) AS ?result)
    FILTER (!isBlank(?result))
} LIMIT 1

For reading purposes only, the above can be loosely rewritten to the following:

SELECT ?result 
{ 
   <http://dbpedia.org/ontology/Village> <http://www.w3.org/2000/01/rdf-schema#subClassOf> ?next. 
   BIND( IF(?next = <http://dbpedia.org/ontology/PopulatedPlace>, 1 , ?next) AS ?result)  # otherwise use recursion
} LIMIT 1