如何将Sparql查询的两个不同但相关的结果集组合为一个？

Question

I am working on a project that finds a correlation between the voting trends in American elections to the demographic data of a state and to their respective voting preferences. I have three separate rdf documents which contain the same keys, but hold different types of data. So I tried the following SPARQL query on Fuseki server,

SELECT ?p ?o1 ?object3
WHERE {
   {  
     ?subject <http://semanticspiders.org/demographic#AZ> ?object .
     ?object <http://semanticspiders.org/demographic#age> ?o .
     ?o ?p ?o1
   } 

  UNION
  {
     ?subject <http://semanticspiders.org/voterTurnout#AZ> ?object .
     ?object <http://semanticspiders.org/voterTurnout#age> ?o .
     ?o ?p ?o1

  }
  UNION 
  {
     ?subject <http://semanticspiders.org/voterBias#age> ?object .
     ?object ?p ?object2 .
     ?object2 <http://semanticspiders.org/voterBias#left> ?object3


  }
  FILTER (?p = <http://semanticspiders.org/demographic#18-29> || ?p = <http://semanticspiders.org/voterBias#18-29>)


}  

which gives me the following result.

However I want result set to look something like the following - {18-29,"21.9","5.161"}

I have tried group by and other filters but couldn't make it work. Any help is appreciated.

Answer 1

1. Delete the two lines that say UNION . Union is for returning rows that match the one pattern or the other pattern. You want a row that matches the one pattern and the other pattern. That's what SPARQL does by default. Keep the curly braces {...} around each group.
2. Delete the FILTER . If you use variables in the right way, you won't need a filter.
3. Make sure that each of the three {...} groups has its own separate variables. They should not share any variables. If you share variables between groups, then they must have the same value or else the row will be removed.
4. By the way, ?subject , ?object and ?p are pretty terrible variable names. Name your variables after the kind of thing that is going to be bound to it. So, ?age or ?voterBias make for better names.
5. Make a new variable that is shared between the three groups, maybe call it ?key , or ?ageRange as proposed in the comments. This variable needs to hold the value that you want to use to join between the groups, so 18-29 for example. So it needs to receive exactly the same value in each of the three groups. It looks like that value doesn't exist directly in your data, so it needs to be computed with an expression and then bound to the variable in each group, using BIND(... AS ?key) .

While doing this, you probably want to work on each of the three groups individually, and only put them all into the query once each group works. Basically, the result for each group needs to have ?key with values that are consistent between the groups, and otherwise only have variables in the result that are not shared with another group.