SPARQL: How can I group multiple fields? - sparql

Question

I'm trying to use the GROUP BY and GROUP_CONCAT functions to group my results into single rows for each entity but I cannot make it work.
This is what I have: a Project has a Name and can have multiple Topics and Areas.
When I do this query:
SELECT ?proj ?name ?topic ?area
WHERE {
?proj hasName ?name .
?proj hasTopic ?topic .
?proj hasArea ?area .
}
I get this table of bindings:
proj
name
topic
area
uri
My Project
building
Europe
uri
My Project
building
Asia
uri
My Project
park
Europe
uri
My Project
park
Asia
But what I would want is:
proj
name
topic
area
uri
My Project
building, park
Europe, Asia
I've tried using:
SELECT ?proj ?name
(GROUP_CONCAT(?topic; separator=", ") AS ?topics)
(GROUP_CONCAT(?area; separator=", ") AS ?areas)
WHERE {
?proj hasName ?name .
?proj hasTopic ?topic .
?proj hasArea ?area .
}
GROUP BY ?proj ?name
But I get the topics doubled: building, building, park, park.
What am I missing? How should I approach this?
Thanks a lot in advance

Answer 1

You are on the right path, but you just need to use the DISTINCT keyword. The following works for me.
SELECT ?proj ?name
(GROUP_CONCAT(DISTINCT ?topic; separator=", ") AS ?topics)
(GROUP_CONCAT(DISTINCT ?area; separator=", ") AS ?areas)
WHERE {
?proj hasName ?name .
?proj hasTopic ?topic .
?proj hasArea ?area .
}
GROUP BY ?proj ?name