wifi-densepose/docs/research/sparql/SPARQL_SPECIFICATION.md

SPARQL 1.1 Specification for PostgreSQL Implementation

Version: SPARQL 1.1 (W3C Recommendation March 2013) Target: PostgreSQL Query Engine Implementation Research Date: December 2025 Project: RuVector-Postgres

---

Table of Contents

1. Introduction
2. Core SPARQL Components
3. SPARQL Syntax
4. SPARQL Operations
5. SPARQL Update
6. Built-in Functions
7. SPARQL Algebra
8. Query Result Formats
9. Implementation Considerations
10. References

---

Introduction

SPARQL (SPARQL Protocol and RDF Query Language) is a W3C standard query language for querying and manipulating RDF (Resource Description Framework) data. RDF is a directed, labeled graph data format representing information as triples (subject, predicate, object).

SPARQL 1.1 Enhancements

SPARQL 1.1 adds significant features over SPARQL 1.0:

• Subqueries: Nested SELECT queries
• Value assignment: BIND and VALUES clauses
• Property paths: Regular expression-like path matching
• Aggregates: COUNT, SUM, AVG, MIN, MAX, GROUP_CONCAT, SAMPLE
• Negation: NOT EXISTS and MINUS operators
• Service federation: Querying remote SPARQL endpoints
• Update operations: INSERT, DELETE, LOAD, CLEAR, etc.

---

Core SPARQL Components

1. RDF Triple Model

The foundation of SPARQL is the RDF triple:

<subject> <predicate> <object>

Components:

• Subject: IRI or blank node
• Predicate: IRI only
• Object: IRI, blank node, or literal

Example:

<http://example.org/person/Alice> <http://xmlns.com/foaf/0.1/knows> <http://example.org/person/Bob>

2. Graph Patterns

Graph patterns are the building blocks of SPARQL queries:

Basic Graph Pattern (BGP)

A set of triple patterns that must all match:

?person foaf:name ?name .
?person foaf:age ?age .

Group Graph Pattern

Multiple patterns enclosed in braces:

{
  ?person foaf:name ?name .
  ?person foaf:age ?age .
  FILTER(?age >= 18)
}

Optional Graph Pattern

Extends solutions with additional patterns if they match:

?person foaf:name ?name .
OPTIONAL { ?person foaf:email ?email }

Semantics: LEFT JOIN - keeps all solutions from the first pattern whether or not the OPTIONAL pattern matches.

Union Graph Pattern

Alternatives - tries multiple patterns:

{
  { ?person foaf:name ?name }
  UNION
  { ?person rdfs:label ?name }
}

Filter Pattern

Constrains solutions with boolean expressions:

?person foaf:age ?age .
FILTER(?age >= 21 && ?age < 65)

3. Query Forms

SPARQL defines four query forms:

SELECT Query

Returns variable bindings as a table:

SELECT ?name ?age
WHERE {
  ?person foaf:name ?name .
  ?person foaf:age ?age .
}

Returns: Solution sequence (table of variable bindings)

CONSTRUCT Query

Builds an RDF graph using a template:

CONSTRUCT {
  ?person ex:hasName ?name .
  ?person ex:hasAge ?age .
}
WHERE {
  ?person foaf:name ?name .
  ?person foaf:age ?age .
}

Returns: RDF graph

Shorthand:

CONSTRUCT WHERE {
  ?s ?p ?o .
}

ASK Query

Returns boolean indicating if pattern matches:

ASK {
  ?person foaf:name "Alice" .
}

Returns: true or false

DESCRIBE Query

Returns RDF description of resources:

DESCRIBE <http://example.org/person/Alice>

Returns: Implementation-specific RDF graph describing the resource

4. Solution Modifiers

Modifiers that transform query results:

ORDER BY

Sorts results by one or more expressions:

SELECT ?name ?age
WHERE { ?person foaf:name ?name ; foaf:age ?age }
ORDER BY DESC(?age) ?name

Options: ASC(?expr) (ascending, default), DESC(?expr) (descending)

DISTINCT

Removes duplicate solutions:

SELECT DISTINCT ?name
WHERE { ?person foaf:name ?name }

REDUCED

Allows (but doesn't require) duplicate elimination:

SELECT REDUCED ?name
WHERE { ?person foaf:name ?name }

LIMIT

Restricts result count:

SELECT ?name
WHERE { ?person foaf:name ?name }
LIMIT 10

OFFSET

Skips initial solutions:

SELECT ?name
WHERE { ?person foaf:name ?name }
OFFSET 20
LIMIT 10

GROUP BY

Groups solutions for aggregation:

SELECT ?company (COUNT(?employee) AS ?empCount)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
}
GROUP BY ?company

HAVING

Filters grouped results:

SELECT ?company (COUNT(?employee) AS ?empCount)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
}
GROUP BY ?company
HAVING (COUNT(?employee) >= 10)

---

SPARQL Syntax

PREFIX Declarations

Associate prefix labels with IRIs:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>

SELECT ?name
WHERE {
  ?person foaf:name ?name .
}

BASE Declaration

Define base IRI for relative IRIs:

BASE <http://example.org/>

SELECT ?name
WHERE {
  <person/Alice> foaf:name ?name .
}

Variable Syntax

Variables start with ? or $:

?name
$age

Note: ?var and $var refer to the same variable.

URI/IRI Syntax

Three ways to specify IRIs:

1. Full IRI: <http://example.org/resource>
2. Prefixed name: prefix:localPart (e.g., foaf:name)
3. Relative IRI: <resource> (resolved against BASE)

Literal Syntax

String Literals

"simple string"
'another string'
"""multi-line
string"""
'''another
multi-line'''

Numeric Literals

42                    # xsd:integer
3.14                  # xsd:decimal
1.5e6                 # xsd:double

Boolean Literals

true
false

Language-Tagged Literals

"chat"@en
"chat"@fr

Typed Literals

"42"^^xsd:integer
"2025-12-09"^^xsd:date
"P1Y2M"^^xsd:duration

Blank Node Syntax

Labeled Blank Nodes

_:label
_:alice

Anonymous Blank Nodes

[]                          # Empty blank node
[ foaf:name "Alice" ]       # Blank node with properties

Blank Node Property Lists

[
  foaf:name "Alice" ;
  foaf:age 30 ;
  foaf:knows [ foaf:name "Bob" ]
]

Triple Pattern Abbreviations

Semicolon (;) - Shared Subject

?person foaf:name "Alice" ;
        foaf:age 30 ;
        foaf:knows ?friend .

# Equivalent to:
?person foaf:name "Alice" .
?person foaf:age 30 .
?person foaf:knows ?friend .

Comma (,) - Shared Subject-Predicate

?person foaf:knows ?bob, ?charlie, ?diana .

# Equivalent to:
?person foaf:knows ?bob .
?person foaf:knows ?charlie .
?person foaf:knows ?diana .

rdf:type Shorthand (a)

?person a foaf:Person .

# Equivalent to:
?person rdf:type foaf:Person .

Collections (RDF Lists)

?list rdf:rest*/rdf:first ?item .

# Or using collection syntax:
?x foaf:knows ( :Alice :Bob :Charlie ) .

---

SPARQL Operations

1. Pattern Matching

Basic Triple Patterns

SELECT ?subject ?object
WHERE {
  ?subject foaf:knows ?object .
}

Multiple Patterns (Conjunction)

SELECT ?name ?email
WHERE {
  ?person foaf:name ?name .
  ?person foaf:email ?email .
}

2. FILTER Expressions

Apply constraints to solutions:

SELECT ?name ?age
WHERE {
  ?person foaf:name ?name .
  ?person foaf:age ?age .
  FILTER(?age >= 18 && ?age < 65)
}

FILTER Operators

Logical:

• && (AND)
• || (OR)
• ! (NOT)

Comparison:

• = (equals)
• != (not equals)
• < (less than)
• > (greater than)
• <= (less than or equal)
• >= (greater than or equal)

Arithmetic:

• + (addition)
• - (subtraction)
• * (multiplication)
• / (division)

Other:

• IN (set membership)
• NOT IN (set non-membership)

3. OPTIONAL Patterns

Left join - augments solutions if pattern matches:

SELECT ?name ?email
WHERE {
  ?person foaf:name ?name .
  OPTIONAL { ?person foaf:email ?email }
}

Multiple OPTIONAL blocks:

SELECT ?name ?email ?phone
WHERE {
  ?person foaf:name ?name .
  OPTIONAL { ?person foaf:email ?email }
  OPTIONAL { ?person foaf:phone ?phone }
}

OPTIONAL with FILTER:

SELECT ?name ?email
WHERE {
  ?person foaf:name ?name .
  OPTIONAL {
    ?person foaf:email ?email .
    FILTER(CONTAINS(?email, "@example.com"))
  }
}

4. UNION Patterns

Disjunction - tries alternative patterns:

SELECT ?name
WHERE {
  {
    ?person foaf:name ?name .
  }
  UNION
  {
    ?person rdfs:label ?name .
  }
}

5. Property Paths

Regular expression-like patterns over properties:

Operators

Operator	Syntax	Description
Sequence	elt1 / elt2	Follow elt1, then elt2
Alternative	elt1 | elt2	Try elt1 or elt2
Inverse	^elt	Reverse direction (object to subject)
Zero or more	elt*	Zero or more occurrences
One or more	elt+	One or more occurrences
Zero or one	elt?	Optional occurrence
Negation	!elt	Not this property
Negated set	!(elt1|elt2)	None of these properties

Examples

Transitive closure:

SELECT ?ancestor
WHERE {
  ?person foaf:knows+ ?ancestor .
}

Path sequence:

SELECT ?grandchild
WHERE {
  ?person ex:hasChild / ex:hasChild ?grandchild .
}

Inverse path:

SELECT ?child
WHERE {
  ?parent ^ex:hasChild ?child .
}

Alternative paths:

SELECT ?name
WHERE {
  ?person (foaf:name | rdfs:label) ?name .
}

Negated property set:

SELECT ?x ?y
WHERE {
  ?x !rdf:type ?y .
}

6. Subqueries

Nested SELECT queries:

SELECT ?name ?avgAge
WHERE {
  {
    SELECT ?company (AVG(?age) AS ?avgAge)
    WHERE {
      ?employee foaf:workplaceHomepage ?company .
      ?employee foaf:age ?age .
    }
    GROUP BY ?company
  }
  ?company rdfs:label ?name .
}

7. Negation

NOT EXISTS

Pattern must not match:

SELECT ?person
WHERE {
  ?person a foaf:Person .
  FILTER NOT EXISTS { ?person foaf:email ?email }
}

MINUS

Set difference operation:

SELECT ?person
WHERE {
  ?person a foaf:Person .
  MINUS { ?person foaf:email ?email }
}

Difference: NOT EXISTS is a filter, MINUS is a set operation.

8. VALUES

Inject inline data:

SELECT ?name ?age
WHERE {
  VALUES (?person ?age) {
    (<http://example.org/alice> 30)
    (<http://example.org/bob> 25)
  }
  ?person foaf:name ?name .
}

Single variable:

VALUES ?x { :a :b :c }

Multiple variables:

VALUES (?x ?y) {
  (:a 1)
  (:b 2)
  UNDEF
}

9. BIND

Assign values to variables:

SELECT ?name ?fullName
WHERE {
  ?person foaf:givenName ?first .
  ?person foaf:familyName ?last .
  BIND(CONCAT(?first, " ", ?last) AS ?fullName)
}

10. Aggregates

Aggregate Functions

• COUNT(?var) or COUNT(*) - Count solutions
• SUM(?expr) - Sum numeric values
• AVG(?expr) - Average numeric values
• MIN(?expr) - Minimum value
• MAX(?expr) - Maximum value
• GROUP_CONCAT(?expr) - Concatenate strings
• SAMPLE(?expr) - Arbitrary value

GROUP BY Example

SELECT ?company (COUNT(?employee) AS ?count) (AVG(?salary) AS ?avgSalary)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
  ?employee ex:salary ?salary .
}
GROUP BY ?company

HAVING Example

SELECT ?company (AVG(?salary) AS ?avgSalary)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
  ?employee ex:salary ?salary .
}
GROUP BY ?company
HAVING (AVG(?salary) > 50000)

11. Named Graphs

Query specific graphs:

SELECT ?name
FROM <http://example.org/graph1>
WHERE {
  ?person foaf:name ?name .
}

GRAPH keyword:

SELECT ?name ?graph
WHERE {
  GRAPH ?graph {
    ?person foaf:name ?name .
  }
}

Multiple graphs:

SELECT ?name
FROM <http://example.org/graph1>
FROM <http://example.org/graph2>
WHERE {
  ?person foaf:name ?name .
}

---

SPARQL Update

SPARQL 1.1 Update provides operations for modifying RDF graphs.

1. INSERT DATA

Add ground triples (no variables):

PREFIX foaf: <http://xmlns.com/foaf/0.1/>

INSERT DATA {
  <http://example.org/alice> foaf:name "Alice" .
  <http://example.org/alice> foaf:age 30 .
}

With named graph:

INSERT DATA {
  GRAPH <http://example.org/graph1> {
    <http://example.org/alice> foaf:name "Alice" .
  }
}

Behavior:

• Creates graph if it doesn't exist (SHOULD)
• Blank nodes are "fresh" (distinct from existing nodes)
• No effect if triples already exist

2. DELETE DATA

Remove specific triples:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>

DELETE DATA {
  <http://example.org/alice> foaf:age 30 .
}

With named graph:

DELETE DATA {
  GRAPH <http://example.org/graph1> {
    <http://example.org/alice> foaf:age 30 .
  }
}

Behavior:

• No error if triples don't exist
• No variables or blank nodes allowed

3. DELETE/INSERT

Pattern-based updates using WHERE clause:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>

DELETE { ?person foaf:age ?age }
INSERT { ?person foaf:age 31 }
WHERE {
  ?person foaf:name "Alice" .
  ?person foaf:age ?age .
  FILTER(?age = 30)
}

Order: DELETE executes before INSERT

Only DELETE:

DELETE { ?person foaf:email ?email }
WHERE {
  ?person foaf:email ?email .
  FILTER(CONTAINS(?email, "@oldcompany.com"))
}

Only INSERT:

INSERT {
  ?person foaf:age 0 .
}
WHERE {
  ?person a foaf:Person .
  FILTER NOT EXISTS { ?person foaf:age ?age }
}

4. DELETE WHERE

Shorthand for DELETE...WHERE without INSERT:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>

DELETE WHERE {
  ?person foaf:email ?email .
  FILTER(CONTAINS(?email, "@spam.com"))
}

5. LOAD

Load RDF document from IRI into graph:

LOAD <http://example.org/data.ttl>

Into named graph:

LOAD <http://example.org/data.ttl> INTO GRAPH <http://example.org/graph1>

Silent mode (no error):

LOAD SILENT <http://example.org/data.ttl>

6. CLEAR

Remove all triples from graph:

CLEAR GRAPH <http://example.org/graph1>

Options:

• CLEAR DEFAULT - Clear default graph
• CLEAR NAMED - Clear all named graphs
• CLEAR ALL - Clear all graphs
• CLEAR SILENT GRAPH <uri> - No error if graph doesn't exist

7. CREATE

Create empty graph:

CREATE GRAPH <http://example.org/graph1>

Silent mode:

CREATE SILENT GRAPH <http://example.org/graph1>

8. DROP

Remove graph entirely:

DROP GRAPH <http://example.org/graph1>

Options:

• DROP DEFAULT - Equivalent to CLEAR DEFAULT
• DROP NAMED - Drop all named graphs
• DROP ALL - Drop all graphs
• DROP SILENT GRAPH <uri> - No error if graph doesn't exist

9. COPY

Copy graph content to another graph:

COPY GRAPH <http://example.org/source> TO GRAPH <http://example.org/dest>

Behavior:

• Destination graph cleared first
• Source unchanged

10. MOVE

Move graph content to another graph:

MOVE GRAPH <http://example.org/source> TO GRAPH <http://example.org/dest>

Behavior:

• Destination graph cleared first
• Source graph cleared after copy

11. ADD

Add graph content to another graph:

ADD GRAPH <http://example.org/source> TO GRAPH <http://example.org/dest>

Behavior:

• Destination graph augmented
• Source unchanged

12. WITH Clause

Default graph for update operations:

WITH <http://example.org/graph1>
DELETE { ?person foaf:age ?age }
INSERT { ?person foaf:age 31 }
WHERE {
  ?person foaf:name "Alice" .
  ?person foaf:age ?age .
}

13. USING and USING NAMED

Specify graphs for WHERE clause:

DELETE { GRAPH <http://example.org/dest> { ?s ?p ?o } }
USING <http://example.org/source>
WHERE {
  ?s ?p ?o .
  FILTER(?p = foaf:age)
}

---

Built-in Functions

1. Logical and Conditional

bound(?var)

Test if variable is bound:

SELECT ?name ?email
WHERE {
  ?person foaf:name ?name .
  OPTIONAL { ?person foaf:email ?email }
  FILTER(bound(?email))
}

IF(?cond, ?then, ?else)

Conditional expression:

SELECT ?name (IF(?age >= 18, "adult", "minor") AS ?status)
WHERE {
  ?person foaf:name ?name .
  ?person foaf:age ?age .
}

COALESCE(?expr1, ?expr2, ...)

Return first non-error value:

SELECT ?name (COALESCE(?email, ?phone, "no contact") AS ?contact)
WHERE {
  ?person foaf:name ?name .
  OPTIONAL { ?person foaf:email ?email }
  OPTIONAL { ?person foaf:phone ?phone }
}

EXISTS { pattern }

Test if pattern matches:

SELECT ?person
WHERE {
  ?person a foaf:Person .
  FILTER EXISTS { ?person foaf:email ?email }
}

NOT EXISTS { pattern }

Test if pattern doesn't match:

SELECT ?person
WHERE {
  ?person a foaf:Person .
  FILTER NOT EXISTS { ?person foaf:email ?email }
}

IN and NOT IN

Set membership:

SELECT ?name
WHERE {
  ?person foaf:name ?name .
  ?person foaf:age ?age .
  FILTER(?age IN (20, 25, 30, 35))
}

2. RDF Term Functions

isIRI(?x) / isURI(?x)

Test if value is IRI:

FILTER(isIRI(?resource))

isBlank(?x)

Test if value is blank node:

FILTER(isBlank(?node))

isLiteral(?x)

Test if value is literal:

FILTER(isLiteral(?value))

isNumeric(?x)

Test if value is numeric:

FILTER(isNumeric(?value))

str(?x)

Convert to string:

SELECT (str(?uri) AS ?string)
WHERE { ?s ?p ?uri }

lang(?x)

Extract language tag:

SELECT ?name (lang(?name) AS ?language)
WHERE {
  ?person foaf:name ?name .
  FILTER(lang(?name) = "en")
}

datatype(?x)

Get datatype IRI:

SELECT ?value (datatype(?value) AS ?type)
WHERE { ?s ?p ?value }

IRI(?x) / URI(?x)

Construct IRI from string:

BIND(IRI(CONCAT("http://example.org/", ?id)) AS ?resource)

BNODE() / BNODE(?label)

Create blank node:

BIND(BNODE() AS ?newNode)

STRDT(?string, ?datatype)

Create typed literal:

BIND(STRDT("42", xsd:integer) AS ?number)

STRLANG(?string, ?language)

Create language-tagged literal:

BIND(STRLANG("hello", "en") AS ?greeting)

UUID()

Generate random UUID:

BIND(UUID() AS ?id)

STRUUID()

Generate string UUID:

BIND(STRUUID() AS ?idString)

3. String Functions

STRLEN(?string)

String length:

SELECT ?name (STRLEN(?name) AS ?length)
WHERE { ?person foaf:name ?name }

SUBSTR(?string, ?start, ?length)

Extract substring:

SELECT (SUBSTR(?name, 1, 3) AS ?initials)
WHERE { ?person foaf:name ?name }

Note: Start position is 1-based

UCASE(?string)

Convert to uppercase:

SELECT (UCASE(?name) AS ?upper)
WHERE { ?person foaf:name ?name }

LCASE(?string)

Convert to lowercase:

SELECT (LCASE(?name) AS ?lower)
WHERE { ?person foaf:name ?name }

STRSTARTS(?string, ?prefix)

Test if string starts with prefix:

FILTER(STRSTARTS(?email, "admin@"))

STRENDS(?string, ?suffix)

Test if string ends with suffix:

FILTER(STRENDS(?email, "@example.com"))

CONTAINS(?string, ?substring)

Test if string contains substring:

FILTER(CONTAINS(?description, "important"))

STRBEFORE(?string, ?search)

Extract text before substring:

SELECT (STRBEFORE(?email, "@") AS ?username)
WHERE { ?person foaf:email ?email }

STRAFTER(?string, ?search)

Extract text after substring:

SELECT (STRAFTER(?email, "@") AS ?domain)
WHERE { ?person foaf:email ?email }

ENCODE_FOR_URI(?string)

Percent-encode for URI:

SELECT (ENCODE_FOR_URI(?name) AS ?encoded)
WHERE { ?person foaf:name ?name }

CONCAT(?string1, ?string2, ...)

Concatenate strings:

SELECT (CONCAT(?first, " ", ?last) AS ?fullName)
WHERE {
  ?person foaf:givenName ?first .
  ?person foaf:familyName ?last .
}

langMatches(?tag, ?range)

Match language tags:

FILTER(langMatches(lang(?label), "en"))

Language ranges:

• "en" - Exact match
• "*" - Any language
• "en-US" - Specific locale

REGEX(?string, ?pattern) / REGEX(?string, ?pattern, ?flags)

Regular expression matching:

FILTER(REGEX(?email, "^[a-z]+@example\\.com$", "i"))

Flags:

• "i" - Case insensitive
• "s" - Dot matches newline
• "m" - Multi-line mode
• "x" - Ignore whitespace

REPLACE(?string, ?pattern, ?replacement) / REPLACE(?string, ?pattern, ?replacement, ?flags)

String replacement:

SELECT (REPLACE(?phone, "[^0-9]", "") AS ?digitsOnly)
WHERE { ?person foaf:phone ?phone }

4. Numeric Functions

abs(?number)

Absolute value:

SELECT (abs(?diff) AS ?absDiff)
WHERE { BIND(?a - ?b AS ?diff) }

round(?number)

Round to nearest integer:

SELECT (round(?value) AS ?rounded)
WHERE { ?item ex:price ?value }

ceil(?number)

Ceiling function:

SELECT (ceil(?value) AS ?ceiling)
WHERE { ?item ex:price ?value }

floor(?number)

Floor function:

SELECT (floor(?value) AS ?floor)
WHERE { ?item ex:price ?value }

RAND()

Random number [0, 1):

SELECT ?item
WHERE {
  ?item a ex:Product .
  FILTER(RAND() < 0.1)
}

5. Date/Time Functions

now()

Current timestamp:

BIND(now() AS ?currentTime)

year(?datetime)

Extract year:

SELECT (year(?date) AS ?year)
WHERE { ?event ex:date ?date }

month(?datetime)

Extract month (1-12):

SELECT (month(?date) AS ?month)
WHERE { ?event ex:date ?date }

day(?datetime)

Extract day (1-31):

SELECT (day(?date) AS ?day)
WHERE { ?event ex:date ?date }

hours(?datetime)

Extract hours (0-23):

SELECT (hours(?timestamp) AS ?hour)
WHERE { ?event ex:timestamp ?timestamp }

minutes(?datetime)

Extract minutes (0-59):

SELECT (minutes(?timestamp) AS ?minute)
WHERE { ?event ex:timestamp ?timestamp }

seconds(?datetime)

Extract seconds (0-59.999...):

SELECT (seconds(?timestamp) AS ?second)
WHERE { ?event ex:timestamp ?timestamp }

timezone(?datetime)

Extract timezone:

SELECT (timezone(?timestamp) AS ?tz)
WHERE { ?event ex:timestamp ?timestamp }

tz(?datetime)

Timezone abbreviation:

SELECT (tz(?timestamp) AS ?tzAbbr)
WHERE { ?event ex:timestamp ?timestamp }

6. Hash Functions

MD5(?string)

MD5 hash (lowercase hex):

SELECT (MD5(?email) AS ?hash)
WHERE { ?person foaf:email ?email }

SHA1(?string)

SHA-1 hash:

SELECT (SHA1(?password) AS ?hash)
WHERE { ?user ex:password ?password }

SHA256(?string)

SHA-256 hash:

SELECT (SHA256(?data) AS ?hash)
WHERE { ?item ex:data ?data }

SHA384(?string)

SHA-384 hash:

SELECT (SHA384(?data) AS ?hash)
WHERE { ?item ex:data ?data }

SHA512(?string)

SHA-512 hash:

SELECT (SHA512(?data) AS ?hash)
WHERE { ?item ex:data ?data }

7. Aggregate Functions

COUNT(?var) / COUNT(*)

Count solutions:

SELECT ?company (COUNT(?employee) AS ?count)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
}
GROUP BY ?company

DISTINCT modifier:

SELECT (COUNT(DISTINCT ?type) AS ?typeCount)
WHERE { ?s rdf:type ?type }

SUM(?expr)

Sum numeric values:

SELECT ?department (SUM(?salary) AS ?totalSalary)
WHERE {
  ?employee ex:department ?department .
  ?employee ex:salary ?salary .
}
GROUP BY ?department

AVG(?expr)

Average numeric values:

SELECT ?department (AVG(?salary) AS ?avgSalary)
WHERE {
  ?employee ex:department ?department .
  ?employee ex:salary ?salary .
}
GROUP BY ?department

MIN(?expr)

Minimum value:

SELECT ?department (MIN(?salary) AS ?minSalary)
WHERE {
  ?employee ex:department ?department .
  ?employee ex:salary ?salary .
}
GROUP BY ?department

MAX(?expr)

Maximum value:

SELECT ?department (MAX(?salary) AS ?maxSalary)
WHERE {
  ?employee ex:department ?department .
  ?employee ex:salary ?salary .
}
GROUP BY ?department

GROUP_CONCAT(?expr) / GROUP_CONCAT(?expr; SEPARATOR = ?sep)

Concatenate values:

SELECT ?person (GROUP_CONCAT(?skill; SEPARATOR = ", ") AS ?skills)
WHERE {
  ?person ex:hasSkill ?skill .
}
GROUP BY ?person

SAMPLE(?expr)

Arbitrary value from group:

SELECT ?company (SAMPLE(?employee) AS ?anyEmployee)
WHERE {
  ?employee foaf:workplaceHomepage ?company .
}
GROUP BY ?company

---

SPARQL Algebra

The SPARQL algebra defines formal semantics for query evaluation.

Core Algebraic Operators

1. Basic Graph Pattern (BGP)

A set of triple patterns:

BGP(tp1, tp2, ..., tpn)

2. Join (⋈)

Combines two graph patterns:

P1 Join P2

Definition: Solutions that match both P1 and P2 with compatible bindings.

3. LeftJoin (⟕)

Left outer join (OPTIONAL):

LeftJoin(P1, P2, expr)

Definition: All solutions from P1, augmented with P2 solutions where compatible and expr is true.

Formal:

LeftJoin(Ω1, Ω2, expr) =
  { merge(μ1, μ2) | μ1 ∈ Ω1, μ2 ∈ Ω2, compatible(μ1, μ2), expr(merge(μ1, μ2)) = true }
  ∪ { μ1 | μ1 ∈ Ω1, ∀μ2 ∈ Ω2: ¬compatible(μ1, μ2) }
  ∪ { μ1 | μ1 ∈ Ω1, ∃μ2 ∈ Ω2: compatible(μ1, μ2), expr(merge(μ1, μ2)) = false }

4. Filter (σ)

Selection operation:

Filter(expr, P)

Definition: Solutions from P where expr evaluates to true.

5. Union (∪)

Disjunction:

Union(P1, P2)

Definition: Solutions from P1 or P2 (bag union).

6. Minus

Set difference:

Minus(P1, P2)

Definition: Solutions from P1 that don't join with any solution from P2.

7. Graph

Named graph pattern:

Graph(g, P)

Definition: Evaluate P against graph g.

8. Extend (Bind)

Add variable binding:

Extend(P, ?var, expr)

Definition: Add binding ?var = expr to each solution in P.

9. Project (π)

Project variables:

Project(P, vars)

Definition: Keep only specified variables from P.

10. Distinct

Remove duplicates:

Distinct(P)

11. Reduced

Allow duplicate removal:

Reduced(P)

12. OrderBy

Sort solutions:

OrderBy(P, conditions)

13. Slice

Limit and offset:

Slice(P, start, length)

14. ToList

Convert to solution sequence:

ToList(P)

Algebraic Properties

Important: Unlike relational algebra, SPARQL's LeftJoin is NOT distributive over Union:

LeftJoin(P1, Union(P2, P3), expr) ≠ Union(LeftJoin(P1, P2, expr), LeftJoin(P1, P3, expr))

This limits algebraic optimization opportunities.

Query Translation

A SPARQL query is translated to algebra in this order:

1. Parse query text to syntax tree
2. Translate patterns to algebra
3. Apply solution modifiers (GROUP BY, ORDER BY, etc.)
4. Apply projection (SELECT variables)
5. Apply slice (LIMIT/OFFSET)

Example:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>

SELECT ?name ?email
WHERE {
  ?person foaf:name ?name .
  OPTIONAL { ?person foaf:email ?email }
}
ORDER BY ?name
LIMIT 10

Translates to:

Slice(
  OrderBy(
    Project(
      LeftJoin(
        BGP(?person foaf:name ?name),
        BGP(?person foaf:email ?email),
        true
      ),
      {?name, ?email}
    ),
    [ASC(?name)]
  ),
  0,
  10
)

---

Query Result Formats

SPARQL supports multiple serialization formats for query results.

1. JSON Format

For SELECT and ASK queries.

SELECT Results

{
  "head": {
    "vars": ["name", "email"]
  },
  "results": {
    "bindings": [
      {
        "name": {
          "type": "literal",
          "value": "Alice"
        },
        "email": {
          "type": "literal",
          "value": "alice@example.com"
        }
      },
      {
        "name": {
          "type": "literal",
          "value": "Bob"
        }
      }
    ]
  }
}

RDF Term Types

IRI:

{
  "type": "uri",
  "value": "http://example.org/alice"
}

Literal:

{
  "type": "literal",
  "value": "Alice"
}

Language-tagged literal:

{
  "type": "literal",
  "value": "Alice",
  "xml:lang": "en"
}

Typed literal:

{
  "type": "literal",
  "value": "42",
  "datatype": "http://www.w3.org/2001/XMLSchema#integer"
}

Blank node:

{
  "type": "bnode",
  "value": "b0"
}

ASK Results

{
  "head": {},
  "boolean": true
}

2. XML Format

SELECT Results

<?xml version="1.0"?>
<sparql xmlns="http://www.w3.org/2005/sparql-results#">
  <head>
    <variable name="name"/>
    <variable name="email"/>
  </head>
  <results>
    <result>
      <binding name="name">
        <literal>Alice</literal>
      </binding>
      <binding name="email">
        <literal>alice@example.com</literal>
      </binding>
    </result>
    <result>
      <binding name="name">
        <literal>Bob</literal>
      </binding>
    </result>
  </results>
</sparql>

RDF Term Elements

IRI:

<uri>http://example.org/alice</uri>

Literal:

<literal>Alice</literal>

Language-tagged:

<literal xml:lang="en">Alice</literal>

Typed literal:

<literal datatype="http://www.w3.org/2001/XMLSchema#integer">42</literal>

Blank node:

<bnode>b0</bnode>

ASK Results

<?xml version="1.0"?>
<sparql xmlns="http://www.w3.org/2005/sparql-results#">
  <head/>
  <boolean>true</boolean>
</sparql>

3. CSV Format

Simplified format without type information:

name,email
Alice,alice@example.com
Bob,

Characteristics:

• Lossy: No type information (IRI vs literal vs blank node)
• Simple: Easy to consume in applications
• Header row: Variable names
• Empty cells: Unbound variables

4. TSV Format

Tab-separated with type encoding:

?name	?email
"Alice"	"alice@example.com"
"Bob"

RDF Term Encoding:

• IRI: <http://example.org/resource>
• Literal: "value"
• Language-tagged: "value"@en
• Typed literal: "value"^^<datatype>
• Blank node: _:label

Characteristics:

• Lossless: Preserves all type information
• SPARQL/Turtle syntax: Uses standard RDF term syntax
• Simple parsing: Split on tabs

---

Implementation Considerations

For PostgreSQL Integration

1. Data Model Mapping

RDF Triples → PostgreSQL Tables:

-- Triple store table
CREATE TABLE rdf_triples (
  id BIGSERIAL PRIMARY KEY,
  subject TEXT NOT NULL,
  subject_type VARCHAR(10) NOT NULL,  -- 'iri', 'bnode'
  predicate TEXT NOT NULL,
  object TEXT NOT NULL,
  object_type VARCHAR(10) NOT NULL,   -- 'iri', 'literal', 'bnode'
  object_datatype TEXT,
  object_language VARCHAR(10),
  graph TEXT
);

-- Indexes for query performance
CREATE INDEX idx_triples_spo ON rdf_triples(subject, predicate, object);
CREATE INDEX idx_triples_pos ON rdf_triples(predicate, object, subject);
CREATE INDEX idx_triples_osp ON rdf_triples(object, subject, predicate);
CREATE INDEX idx_triples_graph ON rdf_triples(graph);

2. Query Translation

SPARQL → SQL Translation:

SPARQL BGP:

?person foaf:name ?name .
?person foaf:age ?age .

SQL Translation:

SELECT t1.subject AS person, t1.object AS name, t2.object AS age
FROM rdf_triples t1
JOIN rdf_triples t2 ON t1.subject = t2.subject
WHERE t1.predicate = 'http://xmlns.com/foaf/0.1/name'
  AND t2.predicate = 'http://xmlns.com/foaf/0.1/age';

3. OPTIONAL → LEFT JOIN

SPARQL:

?person foaf:name ?name .
OPTIONAL { ?person foaf:email ?email }

SQL:

SELECT t1.subject AS person, t1.object AS name, t2.object AS email
FROM rdf_triples t1
LEFT JOIN rdf_triples t2 ON t1.subject = t2.subject
  AND t2.predicate = 'http://xmlns.com/foaf/0.1/email'
WHERE t1.predicate = 'http://xmlns.com/foaf/0.1/name';

4. UNION → UNION ALL

SPARQL:

{ ?person foaf:name ?name }
UNION
{ ?person rdfs:label ?name }

SQL:

SELECT subject AS person, object AS name
FROM rdf_triples
WHERE predicate = 'http://xmlns.com/foaf/0.1/name'
UNION ALL
SELECT subject AS person, object AS name
FROM rdf_triples
WHERE predicate = 'http://www.w3.org/2000/01/rdf-schema#label';

5. FILTER → WHERE

SPARQL:

?person foaf:age ?age .
FILTER(?age >= 18)

SQL:

SELECT subject AS person, object AS age
FROM rdf_triples
WHERE predicate = 'http://xmlns.com/foaf/0.1/age'
  AND object_type = 'literal'
  AND object_datatype = 'http://www.w3.org/2001/XMLSchema#integer'
  AND CAST(object AS INTEGER) >= 18;

6. Property Paths

Property paths require recursive queries:

SPARQL:

?person foaf:knows+ ?ancestor .

SQL (PostgreSQL CTE):

WITH RECURSIVE transitive AS (
  -- Base case
  SELECT subject, object
  FROM rdf_triples
  WHERE predicate = 'http://xmlns.com/foaf/0.1/knows'

  UNION

  -- Recursive case
  SELECT t.subject, r.object
  FROM rdf_triples t
  JOIN transitive r ON t.object = r.subject
  WHERE t.predicate = 'http://xmlns.com/foaf/0.1/knows'
)
SELECT * FROM transitive;

7. Aggregates

SPARQL aggregates map to SQL aggregates:

SPARQL:

SELECT ?company (COUNT(?employee) AS ?count)
WHERE { ?employee foaf:workplaceHomepage ?company }
GROUP BY ?company

SQL:

SELECT object AS company, COUNT(*) AS count
FROM rdf_triples
WHERE predicate = 'http://xmlns.com/foaf/0.1/workplaceHomepage'
GROUP BY object;

8. Optimization Strategies

Statistics-based query planning:

• Collect statistics on predicate frequencies
• Estimate selectivity of triple patterns
• Order joins by selectivity

Materialized views:

• Pre-compute common property paths
• Cache frequently accessed subgraphs

Indexes:

• SPO, POS, OSP indexes (covering all access patterns)
• Partial indexes for specific predicates
• GiST/GIN indexes for full-text search

Caching:

• Query result cache
• Parsed query cache
• Compiled SQL cache

9. PostgreSQL Extensions

Leverage existing PostgreSQL features:

• JSONB: Store complex objects
• Full-text search: Text matching in literals
• GiST indexes: Spatial/hierarchical data
• CTEs: Recursive queries for property paths
• Window functions: Advanced analytics
• Parallel query: Scale to large datasets

10. Integration with RuVector

Combine SPARQL with vector operations:

PREFIX foaf: <http://xmlns.com/foaf/0.1/>
PREFIX rv: <http://ruvector.org/functions/>

SELECT ?name ?similarity
WHERE {
  ?person foaf:name ?name .
  ?person rv:embedding ?embedding .

  # Use RuVector distance function
  BIND(rv:cosine_similarity(?embedding, $query_vector) AS ?similarity)
  FILTER(?similarity > 0.8)
}
ORDER BY DESC(?similarity)
LIMIT 10

Implementation:

SELECT t1.object AS name,
       ruvector_cosine_similarity(
         t2.object::ruvector,
         $1::ruvector
       ) AS similarity
FROM rdf_triples t1
JOIN rdf_triples t2 ON t1.subject = t2.subject
WHERE t1.predicate = 'http://xmlns.com/foaf/0.1/name'
  AND t2.predicate = 'http://ruvector.org/properties/embedding'
  AND ruvector_cosine_similarity(t2.object::ruvector, $1::ruvector) > 0.8
ORDER BY similarity DESC
LIMIT 10;

Performance Considerations

1. Index Strategy: SPO, POS, OSP covering all join orders
2. Query Optimization: Statistics-based join reordering
3. Caching: Parsed queries and compiled SQL
4. Parallelization: Leverage PostgreSQL parallel query
5. Partitioning: By graph, predicate, or subject
6. Connection Pooling: Reuse database connections
7. Prepared Statements: Reduce parsing overhead

Standards Compliance

Implement according to:

• SPARQL 1.1 Query Language (W3C Recommendation)
• SPARQL 1.1 Update (W3C Recommendation)
• SPARQL 1.1 Protocol (HTTP bindings)
• SPARQL 1.1 Results JSON/XML/CSV/TSV Formats

Consider SPARQL 1.2 draft features:

• Enhanced property paths
• New functions
• Improved federation

---

References

W3C Specifications

1. SPARQL 1.1 Query Language - W3C Recommendation, March 2013
2. SPARQL 1.1 Update - W3C Recommendation, March 2013
3. SPARQL 1.1 Protocol - W3C Recommendation, March 2013
4. SPARQL 1.1 Results JSON Format - W3C Recommendation, March 2013
5. SPARQL 1.1 Results CSV/TSV Formats - W3C Recommendation, March 2013
6. SPARQL 1.1 Property Paths - W3C Recommendation, March 2013
7. SPARQL Query Language for RDF (1.0) - W3C Recommendation, January 2008

Draft Specifications

8. SPARQL 1.2 Overview - W3C Working Draft
9. SPARQL 1.2 Query Language - W3C Working Draft

Formal Semantics

10. SPARQL Algebra - W3C Draft
11. ARQ's SPARQL Algebra - W3C Community

Academic Resources

12. Apache Jena SPARQL Tutorials
13. TU Dresden SPARQL Algebra Lectures
14. GraphDB SPARQL Functions Reference

Implementation Guides

15. Virtuoso SPARQL Examples
16. EasyRdf Property Paths

---

Appendix: Grammar Summary

Query Structure

Query ::= Prologue ( SelectQuery | ConstructQuery | DescribeQuery | AskQuery ) ValuesClause
Prologue ::= ( BaseDecl | PrefixDecl )*
BaseDecl ::= 'BASE' IRIREF
PrefixDecl ::= 'PREFIX' PNAME_NS IRIREF

SelectQuery ::= SelectClause DatasetClause* WhereClause SolutionModifier
ConstructQuery ::= 'CONSTRUCT' ( ConstructTemplate DatasetClause* WhereClause | DatasetClause* 'WHERE' '{' TriplesTemplate? '}' ) SolutionModifier
DescribeQuery ::= 'DESCRIBE' ( VarOrIri+ | '*' ) DatasetClause* WhereClause? SolutionModifier
AskQuery ::= 'ASK' DatasetClause* WhereClause SolutionModifier

DatasetClause ::= 'FROM' ( DefaultGraphClause | NamedGraphClause )
WhereClause ::= 'WHERE'? GroupGraphPattern
SolutionModifier ::= GroupClause? HavingClause? OrderClause? LimitOffsetClauses?

GroupClause ::= 'GROUP' 'BY' GroupCondition+
HavingClause ::= 'HAVING' HavingCondition+
OrderClause ::= 'ORDER' 'BY' OrderCondition+
LimitOffsetClauses ::= LimitClause OffsetClause? | OffsetClause LimitClause?
LimitClause ::= 'LIMIT' INTEGER
OffsetClause ::= 'OFFSET' INTEGER

Graph Patterns

GroupGraphPattern ::= '{' ( SubSelect | GroupGraphPatternSub ) '}'
GroupGraphPatternSub ::= TriplesBlock? ( GraphPatternNotTriples '.'? TriplesBlock? )*

GraphPatternNotTriples ::= GroupOrUnionGraphPattern | OptionalGraphPattern | MinusGraphPattern | GraphGraphPattern | ServiceGraphPattern | Filter | Bind | InlineData

OptionalGraphPattern ::= 'OPTIONAL' GroupGraphPattern
GraphGraphPattern ::= 'GRAPH' VarOrIri GroupGraphPattern
ServiceGraphPattern ::= 'SERVICE' 'SILENT'? VarOrIri GroupGraphPattern
Bind ::= 'BIND' '(' Expression 'AS' Var ')'
InlineData ::= 'VALUES' DataBlock
MinusGraphPattern ::= 'MINUS' GroupGraphPattern
GroupOrUnionGraphPattern ::= GroupGraphPattern ( 'UNION' GroupGraphPattern )*
Filter ::= 'FILTER' Constraint

Triple Patterns

TriplesBlock ::= TriplesSameSubjectPath ( '.' TriplesBlock? )?
TriplesSameSubjectPath ::= VarOrTerm PropertyListPathNotEmpty | TriplesNodePath PropertyListPath
PropertyListPath ::= PropertyListPathNotEmpty?
PropertyListPathNotEmpty ::= ( VerbPath | VerbSimple ) ObjectListPath ( ';' ( ( VerbPath | VerbSimple ) ObjectList )? )*
VerbPath ::= Path
VerbSimple ::= Var
ObjectListPath ::= ObjectPath ( ',' ObjectPath )*
ObjectPath ::= GraphNodePath
Path ::= PathAlternative
PathAlternative ::= PathSequence ( '|' PathSequence )*
PathSequence ::= PathEltOrInverse ( '/' PathEltOrInverse )*
PathElt ::= PathPrimary PathMod?
PathEltOrInverse ::= PathElt | '^' PathElt
PathMod ::= '?' | '*' | '+'
PathPrimary ::= iri | 'a' | '!' PathNegatedPropertySet | '(' Path ')'
PathNegatedPropertySet ::= PathOneInPropertySet | '(' ( PathOneInPropertySet ( '|' PathOneInPropertySet )* )? ')'
PathOneInPropertySet ::= iri | 'a' | '^' ( iri | 'a' )

Update Operations

Update ::= Prologue ( Update1 ( ';' Update )? )?
Update1 ::= Load | Clear | Drop | Add | Move | Copy | Create | InsertData | DeleteData | DeleteWhere | Modify

Load ::= 'LOAD' 'SILENT'? iri ( 'INTO' GraphRef )?
Clear ::= 'CLEAR' 'SILENT'? GraphRefAll
Drop ::= 'DROP' 'SILENT'? GraphRefAll
Create ::= 'CREATE' 'SILENT'? GraphRef
Add ::= 'ADD' 'SILENT'? GraphOrDefault 'TO' GraphOrDefault
Move ::= 'MOVE' 'SILENT'? GraphOrDefault 'TO' GraphOrDefault
Copy ::= 'COPY' 'SILENT'? GraphOrDefault 'TO' GraphOrDefault

InsertData ::= 'INSERT DATA' QuadData
DeleteData ::= 'DELETE DATA' QuadData
DeleteWhere ::= 'DELETE WHERE' QuadPattern
Modify ::= ( 'WITH' iri )? ( DeleteClause InsertClause? | InsertClause ) UsingClause* 'WHERE' GroupGraphPattern

DeleteClause ::= 'DELETE' QuadPattern
InsertClause ::= 'INSERT' QuadPattern
UsingClause ::= 'USING' ( iri | 'NAMED' iri )

GraphRefAll ::= GraphRef | 'DEFAULT' | 'NAMED' | 'ALL'
GraphRef ::= 'GRAPH' iri

---

End of Specification Document

This document provides comprehensive coverage of SPARQL 1.1 for implementing a query engine in PostgreSQL. For complete formal definitions and edge cases, refer to the official W3C specifications linked in the References section.