How to exclude nodes from path? - sparql

Question

I want to get all mathematicians from DBpedia, so I wrote this query for DBpedia's SPARQL service:
SELECT DISTINCT ?person
{
?person dct:subject ?category.
?category skos:broader* dbc:Mathematicians.
}
The problem with this is that the category Mathematicians is polluted, due to categories like dbc:Euclid, which then includes all of Euclidean geometry. I believe it's categories like these which cause the query to fail:
Virtuoso 42000 Error TN...: Exceeded 1000000000 bytes in transitive temp memory. use t_distinct, t_max or more T_MAX_memory options to limit the search or increase the pool
A lot of the problematic categories are in dbc:Wikipedia_categories_named_after_mathematicians.
Is there some way to ignore these categories in the skos:broader* path that would make the error go away?

Answer 1

You can list the categories that you don't want to include by filtering them out:
SELECT DISTINCT ?person
{
?person dct:subject ?category.
?category skos:broader* dbc:Mathematicians.
FILTER (?category NOT IN (dbc:Euclid))
}
But that won't remove the error because Virtuoso still needs to traverse the skos:broader hierarchy, exhausting 'transitive heap memory'. Other approaches include selecting specific categories or traversing part of the hierarchy.
The specific category could use UNION statements, but the VALUES shortcut is a simpler syntax:
SELECT DISTINCT ?person
{
VALUES ?category {dbc:Mathematicians dbc:Mental_calculators dbc:Lists_of_mathematicians}
?person dct:subject ?category.
}
For querying part of the hierarchy, you can use some property path expressions. This one will get parents and grandparents:
SELECT DISTINCT ?person
{
?person dct:subject ?category.
?category skos:broader | (skos:broader/skos:broader) dbc:Mathematicians.
# filter as desired - FILTER (?category NOT IN (dbc:Euclid))
}