SPARQL1.1 101 Language and Jena support

1 introduction

definition cited from SPARQL 1.1 Overview:

SPARQL 1.1 is a set of specifications that provide languages and protocols to query and manipulate RDF graph content on the Web or in an RDF store.

I roughly divide W3C's SPARQL 1.1 Specifications into 5 parts:

• language
  SPARQL 1.1 Query Language
  SPARQL 1.1 Update

• protocol
  SPARQL 1.1 Protocol
  SPARQL 1.1 Graph Store HTTP Protocol

• result format
  SPARQL 1.1 Query Results JSON Format
  SPARQL 1.1 Query Results CSV and TSV Formats
  SPARQL Query Results XML Format (Second Edition)

• application
  SPARQL 1.1 Service Description
  SPARQL 1.1 Federated Query
  SPARQL 1.1 Entailment Regimes

• conformant accessment
  SPARQL 1.1 Test Cases

This note is dedicated to record SPARQL languages(query, update), and Jena's ARQ(an implementation) supports.

2 language

No one really want to read the specifications directly, especially the unpatients, including myself. So I choose to stand on the shoulders of giants. Section 2.1 Query is mostly cited from [1].

2.1 Query

At first glance, SPARQL Query's key words seems to be copyed from Relational Storage Query language SQL. Be careful, although the key words may be familiar, the underlying modeling method and storage are different.

2.1.1 SELECT

(1) a sample with projection and modifier:

# prefixes
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>

# select clause, with projection
SELECT ?sub ?prop ?obj
WHERE {
    ?sub ?prop ?obj.
}
LIMIT 0, 10 # using query modifier

The WHERE caluse is a specification of graph pattern, and each line in WHERE caluse is a triple, and should end with .. The notations that begin with ? are variables, and when SAPRQL endpoint generate a query answer, these variables should bind to a specific IRI resource.

(2) a sample cited from "Linked Data in Action" with additional RDF dataset and abbreviation of triple:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX pos: <http://www.w3.org/2003/01/geo/wgs84_pos#>

SELECT ?name ?latitude ?longitude
# named graph
FROM <http://3roundstones.com/dave/me.rdf>
FROM <http://semanticweb.org/wiki/Special:ExportRDF/Michael_Hausenblas>
WHERE {
   ?person foaf:name ?name ;
      foaf:based_near ?near .# abbreviation
   ?near pos:lat ?latitude ;
      pos:long ?longitude .
}

?person foaf:name ?name ;foaf:based_near ?near. uses an abbreviation of turtle syntax. Here are some forms of turtle syntax abbreviations:

a :b c.  ==   a :b c;
a :e f.         :e f.

a :b c.  ==  a :b c;  ==  a :b c,d
a :b d.        :b d.

The FROM caluses define so-called additional RDF dataset, without FROM, the default graph of SPARQL endpoint is used.

(3) queries with named graph and blank nodes

PREFIX tbl: <http://www.w3.org/People/Berners-Lee/card#>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT *
FROM NAMED <http://www.koalie.net/foaf.rdf>
FROM NAMED <http://heddley.com/edd/foaf.rdf>
FROM NAMED <http://www.cs.umd.edu/∼hendler/2003/foaf.rdf>
FROM NAMED <http://www.dajobe.org/foaf.rdf>
FROM NAMED <http://www.isi.edu/∼gil/foaf.rdf>
FROM NAMED <http://www.ivan-herman.net/foaf.rdf>
FROM NAMED <http://www.kjetil.kjernsmo.net/foaf>
FROM NAMED <http://www.lassila.org/ora.rdf>
FROM NAMED <http://www.mindswap.org/2004/owl/mindswappers>
WHERE {
    GRAPH ?originGraph {
        _:blank1 foaf:knows _:blank2.
        _:blank2 rdf:type foaf:Person;
                foaf:nick ?nickname;
                foaf:name ?realname
    }
}

The FROM NAMED <IRI> caluses denote named graphs, they are used to indentify Resource's data sources. The GRAPH ?originGraph {<graph-pattern>} is used to define named graphs' patterns, ?originGraph is called the variable of named graph. Also we can explicitly assign a particular in the GRAPH caluse, here is a example:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT ?nickname
WHERE {
    GRAPH <http://www.w3.org/People/Berners-Lee/card> {
        :blank3 foaf:nick ?nickname
    }
}

Notations like _:blank1 are blank nodes, they play the same role just like Resource variables, but they could not be included in the query answers. Blank nodes has 2 forms of notaions:

_:blank :prop obj.   == [] :prop obj. or
                         [ :prop obj]

_:blank :prop1 obj1. == [:prop1 obj1]
_:blank :prop2 obj2.      :prop2 obj2.

sub :prop1 _:blank.   == sub :prop1 [:prop2 obj].
_:blank :prop2 obj. 

_:blank :prop1 obj1; == [:prop1 obj1;
        :prop2 obj2.       :prop2 obj2].

_:blank :prop obj1,  == [:prop obj1, obj2]
              obj2.

Query Modifiers

(1) DISTINCT

DISTINCT is used to remove duplicated result in query answers.

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbprop: <http://dbpedia.org/property/>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT DISTINCT ?picture
WHERE {
    ?person rdfs:label "George Washington"@en;
        dbprop:occupation ?job;
        dbprop:birthPlace ?birthLoc;
        foaf:img ?picture
}

(2) REDUCED

Just like DISTINCT, excepte that SPARQL endpoint can return any number of duplicated results.

(3) ORDER BY

ORDER BY sort results using alphabetical order and numberic order, and uses ASC() order by default. Use DESC() to specify descented order.

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbprop: <http://dbpedia.org/property/>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT ?job ?birthLoc ?picture
WHERE {
    ?person rdfs:label "George Washington"@en;
        dbprop:occupation ?job;
        dbprop:birthPlace ?birthLoc;
        foaf:img ?picture
}
ORDER BY ?birthLoc DESC(?job)

(4) OFFSET, LIMIT

OFFSET and LIMIT should play with ORDER BY, used to generate streaming results.

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbprop: <http://dbpedia.org/property/>
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT ?job ?birthLoc ?picture
WHERE {
    ?person rdfs:label "George Washington"@en;
        dbprop:occupation ?job;
        dbprop:birthPlace ?birthLoc;
        foaf:img ?picture
}
ORDER BY ?birthLoc DESC(?job)
OFFSET 0 LIMIT 10

(5) FILTER

FILTER is used to filter query answer result through some boolean conditions.

These conditions can be specified using: (a) a subset of XQuery, XPath operators and functions, or (b) SPARQL specific operators.

PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbprop: <http://dbpedia.org/property/>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
SELECT ?prop ?object
WHERE {
    ?person rdfs:label "George Washington"@en;
        dbprop:presidentStart ?start;
        ?prop ?object.
    FILTER(xsd:integer(?start) + 4 <= xsd:integer(?object))
}

The position of FILTER in WHERE clause is not important.

(6) OPTIONAL

OPTIONAL is used to add more graph patterns in WHERE caluses, while does not restrict results when there are no query bindings for these patterns.

PREFIX ex: <http://www.example.com/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbprop: <http://dbpedia.org/property/>
SELECT ?l1 ?l2 ?l3 ?l4
WHERE {
    ?person rdfs:label "George Washington"@en.
    ?l1 dbprop:namedFor ?person.
    OPTIONAL { ?l2 dbprop:blankInfo ?person }
    OPTIONAL { ?l3 ex:isNamedAfter ?person }
    OPTIONAL { ?person ex:wasFamousAndGaveNameTo ?l4 }
}

(7) UNION

UNION is used to aggregate results of two or more graph patterns to generate a new result.

PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT *
WHERE {
    { ?unknown foaf:gender "male" }
    UNION
    { ?unknown foaf:gender "female" } .

    { ?unknown foaf:interest <http://www.iamhuman.com> }
    UNION
    { ?unknown foaf:interest <http://lovebeinghuman.org> }
}

CAUTION: No examples are expressed in CONSTRUCT, ASK and DESCRIBE.

2.1.2 CONSTRUCT

CONSTRUCT is used to transform datas from RDF datasets to datas in another RDF dataset.

2.1.3 ASK

ASK is used when you want to confirm whether a given graph pattern is exist in the SPARQL endpoint, it returns a boolean result.

2.1.4 DESCRIBE

DESCRIBE is used when query clients do not know the details of the structure of data, the SPARQL endpoint decides how to organize the data.

2.2 Update

TODO

3 language implementation: Jena ARQ

Using Jena with Apache Maven:

<groupId>org.apache.jena</groupId>
<artifactId>apache-jena-libs</artifactId>
<type>pom</type>
<version>2.13.0</version>
</dependency>

3.1 API

3.1.1 query from model

see SelectQueryUsingModel.java

3.1.2 query from dataset

see SelectQueryUsingDataset.java

or paly with existed models:

Dataset dataset = DatasetFactory.create() ;
dataset.setDefaultModel(model) ;
dataset.addNamedModel("http://example/named-1", modelX) ;
dataset.addNamedModel("http://example/named-2", modelY) ;
 try(QueryExecution qExec = QueryExecutionFactory.create(query, dataset)) {
    ...
}

3.1.3 query from remote service

see SelectQueryUsingRemoteService.java

3.1.4 store resultset for later use

ResultSet results = qexec.execSelect() ;
results = ResultSetFactory.copyResults(results) ;
return results;

3.2 command line utilities

Classes are located in package arq.

(1) run a remote service query

java -cp .. arq.rsparql --service 'http://www.sparql.org/books/sparql'  \
   'PREFIX books:   <http://example.org/book/>  \
    PREFIX dc:      <http://purl.org/dc/elements/1.1/>  \
       SELECT ?book ?title WHERE   { ?book dc:title ?title }'

(2) run a remote service query with a local query file

java -cp .. arq.rsparql --service='http://www.sparql.org/books/sparql' \
 --file=query/books.rq

here query directory specified in --file arguments is in the CLASSPATH.

Other command line classes refer ARQ - Command Line Applications for more details.

(3) run with an OS script

cd $JENA_HOME/bin
./rsparql --service 'http://www.sparql.org/books/sparql' \
     'PREFIX books:   <http://example.org/book/> \
      PREFIX dc:      <http://purl.org/dc/elements/1.1/> \
      SELECT ?book ?title WHERE   { ?book dc:title ?title }'

# or
./rsparql --service='http://www.sparql.org/books/sparql' --file=./query/books.rq

3.3 other queries except SELECT

CONSTRUCT

QueryExecution qexec = ...;
Model resultModel = qexec.execConstruct() ;

DESCRIBE

QueryExecution qexec = ...;
Model resultModel = qexec.execDescribe() ;

ASK

QueryExecution qexec = ...;
boolean result = qexec.execAsk() 

resources and references

[1] Hebeler J, Fisher M, et al.Web 3.0与Semantic Web编程[M]. 清华大学出版社, 北京.2010

[2] SPARQLer - SPARQL online tools and resources collection