1. Logics for Data and Knowledge Representation SPARQL Protocol and RDF Query Language (SPARQL) Feroz Farazi

2. SPARQL  A language for expressing queries to retrieve information from various data represented in RDF [SPARQL Spec.]  A query language with the capability to search graph patterns [SPARQL Spec.]  Often SPARQL queries contain  a basic graph pattern: a set of subject, object, predicate triple patterns  RDF terms possibly substituted with variables  Result of the query  a subgraph of the RDF data graph

3. Terminologies  RDF Terms: Given that I is the set of all IRIs, L is the set of all RDF literals and B is the set of all blank nodes in an RDF graph. Within the graph the set of all RDF Terms, T = I U L U B  RDF Dataset: D = {G, (I 1, G 1 ), (I 2, G 2 ),…(I k, G k )}, where G is the default graph (I i, G i ) are named graphs, i = 1 to k, k ≥ 0  An RDF dataset always contains a default graph, which does not have a name  It contains zero or more named graphs  Each named graph is identified by an IRI

4. Terminologies  Query Variable: A query variable, v ∈ V, where V is infinite and V ∩ T = ∅  Triple Pattern: A triple pattern P ∈ {(T U V) x (I U V) x (T U V)}  Solution Mapping: A solution mapping is a partial function M:V -> T where V is the query variable and T is the set of all RDF Terms  Solution Sequence: A list of solutions which might be unordered. Number of solutions might be zero, one or more.

5. Terminologies  Solution Sequence Modifier: (i) Order By (ii) Projection (iii) Distinct (iv) Reduced (v) Offset (vi) Limit  Others: IRI (Internationalized Resource Identifier), Lexical form, language tag (e.g., en, it), datatype IRI (e.g., xsd:boolean), literal, plain literal and typed literal  IRIs and URIs  URIs include a subset of the ASCII character set  IRIs can include Unicode characters (Universal Character Set)  ASK: to perform a test to know if a query expression has a solution. It replies with yes/no.

6. Query  Dataset: :paper1:title“Semantic Matching” Query Expression: SELECT ?title WHERE { :paper1 :title ?title. } Query Result: “Semantic Matching” title  Dataset: :paper1:creator“Fausto Giunchiglia” Query Expression: SELECT ?author WHERE { :paper1 :creator ?author. } Query Result: “Fausto Giunchiglia” SELECT query form returns RDF Terms bound to the variables

7. Query  Dataset: _:a :name "Tim Berners-Lee". _:a :homepage. _:b :name "Fausto Giunchiglia". _:b :homepage. Query Expression: SELECT ?name ?homepage WHERE { ?x :name ?name. ?x :homepage ?homepage } Query Result: Multiple Matches namehomepage Tim Berners-Lee Fausto Giunchiglia

8. Query  Dataset: :x :name "Tim Berners-Lee"@en. :y :name "Fausto Giunchiglia"@en. Query Expression 1: SELECT ?u WHERE { ?u :name "Tim Berners-Lee"} Query Result: u Query Expression 2: SELECT ?u WHERE { ?u :name "Tim Berners-Lee"@en} Query Result: u :x RDF Literals Matching This query has 0 solution because without language tag the search element does not match with dataset element This query has 1 solution because the inclusion of language tag bound u to :x

9. Building RDF Graphs  CONSTRUCT: this query construct returns an RDF graph  Dataset: _:a :creator "Tim Berners-Lee". _:b :creator "Fausto Giunchiglia". Query Expression: CONSTRUCT { ?x :name ?name } WHERE { ?x :creator ?name } Query Result: _:c :name "Tim Berners-Lee". _:d :name "Fausto Giunchiglia".  In this dataset with :creator we mean Dublin Core (dc) creator metadata  In the query with :name we mean FOAF name metadata  We built a graph with FOAF name attribute which was not available in the source dataset

10. RDF Term Restrictions  FILTER: solutions are restricted to those RDF Terms which match with the filter expression  Dataset: _:a :creator "Tim Berners-Lee". _:a :age 53. _:b :creator "Fausto Giunchiglia". _:b :age 54. Query Expression: SELECT ?author WHERE { ?x :creator ?author. FILTER regex(?author, "Tim") } Query Result: author "Tim Berners-Lee".  The above query can be made case insensitive by adding “i” flag in the filter as follows: FILTER regex(?author, “tim”, “i”) Query Expression: SELECT ?author ?age WHERE { ?x :creator ?author. ?x :age ?age FILTER (?age >53) } Query Result: authorage "Fausto Giunchiglia" 54

11. Querying Optional Pattern  OPTIONAL: to allow binding variables to RDF Terms to be included in the solution in case of availability  Dataset: _:a :creator "Tim Berners-Lee". _:a :age 53. _:a :homepage. _:b :creator "Fausto Giunchiglia". _:b :age 54. Query Expression: SELECT ?author ?homepage WHERE { ?x :creator ?author. OPTIONAL {?x :homepage ?homepage}} Query Result: author homepage "Tim Berners-Lee" "Fausto Giunchiglia"  It is a left associative operator  Why do we need it? All entities might not have the same set of attributes

12. ORDER BY Clause  ORDER BY: a facility to order a solution sequence  Dataset: _:a :creator "Tim Berners-Lee". _:a :age 53. _:b :creator "Fausto Giunchiglia". _:b :age 54. Query Expression: SELECT ?author WHERE { ?x :creator ?author; ?x :age ?age} ORDER BY ?author DESC (?age) Query Result: author "Fausto Giunchiglia" "Tim Berners-Lee"

13. DISTINCT and REDUCED Modifiers  DISTINCT: to remove duplicate from a solution sequence Dataset: _:b :creator "Fausto Giunchiglia". _:b :age 54. _:c :creator "Fausto Giunchiglia". _:c :age 54. Query Expression: SELECT DISTINCT ?creator WHERE { ?x :creator ?creator} Query Result: creator "Fausto Giunchiglia"  REDUCED: to permit the duplicates to be removed. Query Expression: SELECT REDUCED ?creator WHERE { ?x :creator ?creator} The cardinality of the elements in the solution set is at least one and no more than the cardinality without removing duplicates

14. OFFSET and LIMIT Clauses  OFFSET: to show the elements of the solution set starting after a specified number. If the number is zero, there will be no effect. Dataset: _:b :creator "Fausto Giunchiglia". _:b :age 54. _:c :creator "Tim Berners-Lee". _:c :age 53. Query Expression: SELECT ?author WHERE { ?x :creator ?author } ORDER BY ?author OFFSET 1 Query Result: author "Tim Berners-Lee"  Limit: to put an upper bound on the number of elements of the solution set returned Query Expression: SELECT ?author WHERE { ?x :creator ?author } ORDER BY ?author LIMIT 1 OFFSET 1 Query Result: author "Tim Berners-Lee"

15. Relational vs RDF queries  Relational queries consist of (among others):  Relational algebra of joins  Foreign key references  RDF queries consists of (among others):  (Logical) statements in triple form  Unification variables are used to connect graph patterns  A relational query:  Produces a new database table that is a combination of two or more input tables (partially or completely)  An RDF query:  Produces a subset of the input RDF graph  Simplifies some issues of table based queries, for example, no need to put subquery construct [D. Allemang and J. Hendler, 2008]

16. Turtle  Turtle is a terse RDF triple language  Turtle is a textual syntax for RDF facilitates writing RDF graph  in a compact and natural language text form  with abbreviations for common usage patterns and datatypes  compatible with triple pattern syntax of SPARQL (and N-Triples)  Triple:  a sequence of (subject, predicate, object) terms  separated by whitespace  terminated by '.' after each triple e.g.,.  List of predicates and objects:  For the same subject can be codified without repeating the common part  Reference to the subject of the previous triple is indicated by the use of a semicolon e.g., ; "Weaving the Web".

17. Turtle  List of objects:  For the same subject and predicate, triples can be codified without repeating the common part  Reference to the subject and predicate of the previous triple is indicated by the use of a comma e.g., "Tim Berners-Lee", "TBL", "Tim BL".  SPARQL differs from Turtle:  SPARQL permits RDF Literals as the subject of RDF triples  SPARQL permits variables (?name or $name) in any part of the triple of the form  prefix and base declarations  Turtle allows prefix and base declarations anywhere outside of a triple  In SPARQL, they are only allowed in the Prologue (at the start of the SPARQL query)  case sensitivity  SPARQL uses case insensitive keywords, except for 'a‘, where 'a' means the IRI http://www.w3.org/1999/02/22-rdf-syntax-ns#type  Turtle's prefix and base declarations are case sensitive  'true' and 'false' are case insensitive in SPARQL and case sensitive in Turtle  TrUe is not a valid boolean value in Turtle

18. SPARQL: Federated Query  Federated SPARQL query can be used to express queries across diverse data sources if  data is stored natively as RDF or  data is viewed as RDF via middleware  It is an opportunity for data consumers to get data distributed across the Web  Federated Query is used for executing queries distributed over different SPARQL endpoints  The SERVICE keyword  allows a query author to direct a portion of a query to a particular SPARQL endpoint  supports SPARQL queries merging data distributed across the Web

19. SPARQL: Federated Query  An example query to a remote SPARQL endpoint  Consider a query to find the names of the people we know  Data about the names of various people is available at the http://people.example.org/sparql endpoint: and one wants to combine with a local FOAF file http://example.org/myfoaf.rdf that contains the single triple: QUERY: PREFIX foaf: SELECT ?name FROM WHERE { foaf:knows ?person. SERVICE { ?person foaf:name ?name. } } This query, on the datasets provided above, has one solution: RESULT: Name ---------- James DATASET 1: @prefix foaf:. @prefix :. :people1 foaf:name "Tim BL". :people2 foaf:name "James". DATASET 2:.

20. SPARQL: Federated Query  An example query with OPTIONAL to two remote SPARQL endpoints  Consider we want to query people and optionally obtain their interests and the names of people they know  Data in the default graph at remote SPARQL endpoint: At http://people.example.org/sparql DATASET 1: @prefix foaf:. @prefix :. :people1 foaf:name "Tim BL". :people2 foaf:name "James". :people3 foaf:name "Jerome". :people3 foaf:interest. At http://people2.example.org/sparql DATASET 2: @prefix foaf:. @prefix :. :people1 foaf:knows :people21. :people21 foaf:name " Chris". :people3 foaf:knows :people22. :people22 foaf:name “Frank". QUERY: PREFIX foaf: SELECT ?person ?interest ?known WHERE { SERVICE { ?person foaf:name ?name. OPTIONAL { ?person foaf:interest ?interest. SERVICE { ?person foaf:knows ?known. } } This query, on the datasets provided above, has three solutions: RESULT: person interest known --------------------------------------------------------- Tim BL James Jerome

21. SPARQL Update  Graph Store: an editable repository of RDF graphs managed by a single service  Update service: a service (often referred to by the informal term SPARQL endpoint) that accepts and processes update requests  Similarly to RDF dataset a Graph Store contains one (unnamed) slot holding a default graph and zero or more named slots holding named graphs  Formal definition: Graph Store The Graph Store can be viewed as a mutable RDF Dataset. GS = {DG, (iri 1, G 1 ),..., (iri n, G n ) }  Where the default graph DG is the RDF graph associated with the unnamed slot  n ≥ 0 and for each 1 ≤ i ≤ n, G i is an RDF graph associated with the named slot identified by IRI iri i  all IRIs are distinct, i.e., i≠j implies iri i ≠iri j

22. SPARQL Update  Update operations can specify the named graph(s) to be edited. In case the named graph is not mentioned, the operation is performed on the default graph  The unnamed or default graph may refer to a separate graph, a graph describing the named graphs, a representation of a union of other graphs, and so on  Unlike an RDF Dataset, named graphs can be added to or deleted from a Graph Store  A Graph Store can keep local copies of RDF graphs defined elsewhere on the Web and modify those copies independently of the original graph  SPARQL Update supports two categories of update operations on a Graph Store  Graph Update - addition and removal of triples from some graphs  Graph Management - creation and deletion of graphs and the graph update operations (to add, move, and copy graphs) for managing graphs  In the case where there is one unnamed graph and no named graphs, SPARQL Update can be used  as a graph update language (as opposed to a Graph Store update language)

23. References  SPARQL Spec. (2008). W3C Recommendation.  SPARQL 1.1 Federated Query (2013). W3C Recommendation.  SPARQL 1.1 Update (2013). W3C Recommendation.  Turtle Terse RDF Triple Language (2013). W3C Recommendation.  D. Allemang and J. Hendler. Semantic web for the working ontologist: modeling in RDF, RDFS and OWL. Morgan Kaufmann Elsevier, Amsterdam, NL, 2008.