Protégé: Différent results between DL Query and SPARQL Query

Question

Here is a small ontology called wildlife.owl, created by Protégé, in which I have classes animal, carnivore, herbivore, lion, giraffe, and individuals Léo (a lion), Gigi (a giraffe) and Giginou (also a giraffe). In the ontology I only declare that lion ⊏ carnivore ⊏ animal.

When I ask for the instances of animal in the DL Query tab of Protégé, I get, among others, Léo (which is a lion, hence a carnivore, hence an animal).

But when I write the following SPARQ Query:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>
PREFIX me: <file:wildlife.owl#>
SELECT ?b
    WHERE { ?b rdf:type me:animal }

I don't get any instance. Same result when I replace me:animal by me:carnivore. Only when I replace it by me:lion I get the desired result Léo.

Why is DL Query doing the inference (allowing me to obtain Léo as instance of the animal class) and not SPARQL Query?

What can I do to get the same result in SPARQL Query?

Thanks to the answer by @UninformedUser I now know that I must use the Snap SPARQL Query and not the SPARQL Query.

My next question concerns Python: when I send the SPARQL Query using Owlready2 and RDFlib, again I get no result:

from owlready2 import *
from rdflib import *
onto = get_ontology("wildlife.owl").load()
sync_reasoner([onto])
graph = default_world.as_rdflib_graph()
print(list(graph.query_owlready("""
PREFIX rdf-syntax: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX me: <file:wildlife.owl#>
SELECT ?b WHERE {
?b rdf-syntax:type me:animal .
}""")))

How can I get this query use OWL Reasoner?

DL Query Tab uses the OWL reasoner (if enabled), the SPARQL tab not. Snap SPARQL plugin does also use the OWL reasoner, but has some limitation w.r.t. SPARQL syntax. — UninformedUser, Mar 22 '20 at 15:46
I tried `sync_reasoner()` in the Python Owlreader2 code but nothing changed. — yannis, Mar 22 '20 at 18:29
Within the Python's RDFlib collection of RDF tools is [OWL-RL](https://github.com/RDFLib/OWL-RL) which can do RDFS or RL reasoning but not DL. — Nicholas Car, Mar 23 '20 at 00:39
The API is called "owlready2" and not "Owlreader2" - please always check your text before submitting it, otherwise things get confusing. — UninformedUser, Mar 23 '20 at 07:54
And not that it matters, but why the heck are you writing `rdf-syntax:` - everybody is writing `rdf` - please try to stick to common standards. — UninformedUser, Mar 23 '20 at 07:59
Regarding your question: I never used owlready2 in combination with rdflib. Have to try it first. — UninformedUser, Mar 23 '20 at 08:00

score 2 · Accepted Answer · edited Mar 23 '20 at 12:22

When calling the reasoner, Owlready does not retain trivia inferences, such as is-a transitivity (e.g. the fact that lion is an animal).

For trivial inferences, you should rather use SPARQL, as in the example below. The ?any_animal variable contains all animal subclasses (including animal itself), thank to the SubclassOf* SPARQL syntax (the * indicates transitivity). Then, we take any instance of ?any_animal class.

    from owlready2 import *
    from rdflib import *

    onto = get_ontology("http://test.org/wildlife.owl")

    with onto:
        class animal(Thing): pass
        class carnivore(animal): pass
        class lion(carnivore): pass

        lion()

    default_world.graph.dump()

    graph = default_world.as_rdflib_graph()

    print(list(graph.query_owlready("""
    PREFIX rdf-syntax: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
    PREFIX me: <http://test.org/wildlife.owl#>
    SELECT ?b WHERE {
    ?any_animal <http://www.w3.org/2000/01/rdf-schema#subClassOf>* me:animal .
    ?b rdf-syntax:type ?any_animal .
    }""")))

Thank a lot! I have still a question: among the results I get `[[wildlife.Léo], [wildlife.lion1], [wildlife.Gigi], [wildlife.Giginou]]`, as far as I know there is no individual called `wildlife.lion1`, where does this come from? — yannis, Mar 23 '20 at 09:49
I my example, lion1 comes from "lion()". When creating individuals, if their IRI is not explicitly given, Owlready generates a new IRI by taking the class IRI in lower case, and add a number (starting at 1 as here). — Jiba, Mar 24 '20 at 13:14

Protégé: Différent results between DL Query and SPARQL Query

1 Answers1