Cheatsheets:SPARQL

Quick reference for smart people — part of our dev cheatsheets collection.

SPARQL 1.1 quick syntax reference — for people who already know what SPARQL is and need reminders on syntax. For a proper tutorial, take the [https://www.wikidata.org/wiki/Wikidata:SPARQL_tutorial Wikidata SPARQL tutorial] (works against any Wikibase endpoint). For querying *this* instance (endpoint, prefixes, label service), see Help:Contributing/query.

Query forms

Form	Returns	Example
SELECT	table of variable bindings	SELECT ?s ?p ?o WHERE { ?s ?p ?o }
ASK	boolean — does the pattern match?	ASK WHERE { ?s ?p ?o }
CONSTRUCT	an RDF graph	CONSTRUCT { ?s ?p ?o } WHERE { ?s ?p ?o }
DESCRIBE	a graph describing the resource	DESCRIBE ?s WHERE { ?s ?p ?o }

Prefixes & triple patterns

Prefixes abbreviate IRIs; the binding is endpoint-defined.

PREFIX ex: <http://example.org/>
SELECT ?s ?p ?o WHERE {
  ?s ?p ?o .                    # triple pattern: subject predicate object
  FILTER(?p = ex:age)           # constraint on a variable
}

Anonymous/blank node as subject or object: [] ex:age 42. Literals can carry datatypes or language tags: "text"^^xsd:string, "café"@fr.

Solution modifiers

Clause	Effect
DISTINCT	drop duplicate solution rows
ORDER BY	sort: ORDER BY ASC(?x) DESC(?y)
LIMIT n / OFFSET n	page through results
GROUP BY	group rows for aggregation (see below)
HAVING	filter groups, like WHERE for aggregates

FILTER expressions

Category	Examples
comparison	= < > <= >= !=
logical	&& || !
string	STR() CONTAINS() STRSTARTS() STRENDS() REGEX()
numeric	ABS() ROUND() FLOOR() CEIL() RAND()
date/time	YEAR() MONTH() DAY() NOW()
terms	isIRI() isBlank() isLiteral() LANG() DATATYPE()

FILTER(CONTAINS(STR(?label), "cat"))
FILTER(LANG(?label) = "fr")

VALUES & BIND

VALUES ?p { ex:age ex:name }        # restrict variable to a list
BIND(?a + ?b AS ?sum)               # compute and bind a new variable

OPTIONAL, UNION, MINUS

?s ex:name ?n .
OPTIONAL { ?s ex:age ?a }           # left join — ?a missing when absent
{ ?s ex:age ?a } UNION { ?s ex:name ?n }   # union of two patterns
?s ?p ?o .
MINUS { ?s ex:age 42 }              # remove matches

Gotcha: FILTER NOT EXISTS vs

MINUS

differ when a variable is

unbound — prefer MINUS for set-difference semantics.

Property paths

Path	Meaning
ex:parent/ex:parent	sequence (length 2)
ex:parent+	one or more
ex:parent*	zero or more
ex:parent?	zero or one
ex:p1|ex:p2	either property
^ex:parent	inverse direction
!ex:p1	any property except p1

Aggregates

SELECT ?p (COUNT(?s) AS ?n) (SUM(?x) AS ?total) (SAMPLE(?o) AS ?any)
WHERE { ?s ?p ?o . ?s ex:val ?x }
GROUP BY ?p
HAVING (COUNT(?s) > 2)

Other aggregates: AVG() MIN() MAX(), GROUP_CONCAT(?o; SEPARATOR=", ").

Subqueries

SELECT ?s WHERE {
  { SELECT ?s (MAX(?v) AS ?maxv) WHERE { ?s ex:val ?v } GROUP BY ?s }
  ?s ex:val ?maxv
}

Federated queries (SERVICE)

SELECT ?label WHERE {
  ?s ex:ref ?w .
  SERVICE <https://www.wikidata.org/sparql> {
    ?w rdfs:label ?label . FILTER(LANG(?label) = "en")
  }
}

Common tasks

• Count rows: SELECT (COUNT(*) AS ?n) WHERE { ?s ?p ?o }
• Deduplicate: add DISTINCT
• Reverse direction: ?child ^ex:parent ?parent
• Check existence: ASK
• JSON output: append &format=json to the endpoint URL

Gotchas

• The default is AND (join), not OR — use UNION for alternatives.
• An unbound variable in

  FILTER
  

  makes the row fail (filter is not "true" for unbound) — guard with

  BOUND()
  

  or use OPTIONAL.
• Variables used only inside a property path (e.g.

  ?s ex:p1/ex:p2 ?o
  

  ) cannot be selected.
• Blank-node labels (e.g.

  _:x
  

  ) are local to one query — they are not IRIs.
• Aggregates require GROUP BY for non-aggregated variables; forgetting it mixes unrelated rows.

• LIMIT 0
  

  returns no rows but still validates the query.

Further reading

• Wikidata SPARQL tutorial — the recommended tutorial
• SPARQL 1.1 Query Language — official spec
• Help:Contributing/query — querying this instance (endpoint, prefixes, label service)