Querying Wikidata with SPARQL

SPARQL is a query language for retrieving data from knowledge graphs like Wikidata. If you’ve used SQL to query databases, SPARQL follows a similar logic – but instead of rows and columns, you’re working with linked triples: subject, predicate, object.

For this project, we used SPARQL to pull structured data about Black artists in Philadelphia from Wikidata’s public endpoint at query.wikidata.org.

A Basic Query

Here’s a minimal query that retrieves artists tagged with the PMA’s African American artist entity (Q94124522):

SELECT ?artist ?artistLabel WHERE {
  ?artist wdt:P5008 wd:Q94124522 .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "en" .
  }
}

• wdt:P5008 is the “on focus list of Wikimedia project” property
• wd:Q94124522 is the PMA African American artist maintenance category
• The SERVICE wikibase:label block tells Wikidata to return human-readable labels instead of just Q-numbers

Adding More Fields

The real power of SPARQL is pulling multiple properties at once. Our project queries include gender, birth/death dates, birthplace, education, occupation, employer, residence, and which museum collections hold their work. Each OPTIONAL clause adds a field without excluding artists who lack that data:

OPTIONAL { ?artist wdt:P21 ?sexGender. }
OPTIONAL { ?artist wdt:P569 ?birthday. }
OPTIONAL { ?artist wdt:P19 ?birthPlace. }
OPTIONAL { ?artist wdt:P106 ?occupation. }
OPTIONAL { ?artist wdt:P6379 ?worksInCollection. }

Handling Duplicates

One challenge: artists with multiple occupations or residences produce multiple rows. We handle this in SPARQL using GROUP_CONCAT to collapse multiple values into comma-separated strings:

(GROUP_CONCAT(DISTINCT(?occupationLabel); separator=", ") AS ?occupations)

Try It Yourself

You can run SPARQL queries directly in the browser at query.wikidata.org. Our full query notebooks are available in the wikidata repository.

For more background on SPARQL and Wikidata, see the blog posts on the Scholars Studio site.

• Sparql