OWLAPY is a Python Framework for creating and manipulating OWL Ontologies.
Have a look at the Documentation.
git clone https://github.com/dice-group/owlapy
conda create -n temp_owlapy python=3.10.13 --no-default-packages && conda activate temp_owlapy && pip3 install -e .
or
pip3 install owlapy
# To download RDF knowledge graphs
wget https://files.dice-research.org/projects/Ontolearn/KGs.zip -O ./KGs.zip && unzip KGs.zip
pytest -p no:warnings -x # Running 103 tests
Click me!
from owlapy.class_expression import OWLClass, OWLObjectIntersectionOf, OWLObjectSomeValuesFrom
from owlapy.owl_property import OWLObjectProperty
from owlapy import owl_expression_to_sparql, owl_expression_to_dl
from owlapy.owl_ontology_manager import OntologyManager
from owlapy.owl_axiom import OWLDeclarationAxiom, OWLClassAssertionAxiom
from owlapy.owl_individual import OWLNamedIndividual, IRI
# Using owl classes to create a complex class expression
male = OWLClass("http://example.com/society#male")
hasChild = OWLObjectProperty("http://example.com/society#hasChild")
hasChild_male = OWLObjectSomeValuesFrom(hasChild, male)
teacher = OWLClass("http://example.com/society#teacher")
teacher_that_hasChild_male = OWLObjectIntersectionOf([hasChild_male, teacher])
# You can render and print owl class expressions in Description Logics syntax or convert it to SPARQL for example.
print(owl_expression_to_dl(teacher_that_hasChild_male)) # (∃ hasChild.male) ⊓ teacher
print(owl_expression_to_sparql(teacher_that_hasChild_male)) # SELECT DISTINCT ?x WHERE { ?x <http://example.com/society#hasChild> ?s_1 . ?s_1 a <http://example.com/society#male> . ?x a <http://example.com/society#teacher> . } }
# Create an Ontology, add the axioms and save the Ontology.
manager = OntologyManager()
new_iri = IRI.create("file:/example_ontology.owl")
ontology = manager.create_ontology(new_iri)
john = OWLNamedIndividual("http://example.com/society#john")
male_declaration_axiom = OWLDeclarationAxiom(male)
hasChild_declaration_axiom = OWLDeclarationAxiom(hasChild)
john_declaration_axiom = OWLDeclarationAxiom(john)
john_a_male_assertion_axiom = OWLClassAssertionAxiom(john, male)
ontology.add_axiom([male_declaration_axiom, hasChild_declaration_axiom, john_declaration_axiom, john_a_male_assertion_axiom])
ontology.save()
Every OWL object that can be used to classify individuals, is considered a class expression and inherits from OWLClassExpression class. In the above examples we have introduced 3 types of class expressions:
Like we showed in this example, you can create all kinds of class expressions using the OWL objects in owlapy api.
Click me!
from owlapy.owl_ontology_manager import OntologyManager
from owlapy.owl_reasoner import SyncReasoner
from owlapy.static_funcs import stopJVM
ontology_path = "KGs/Family/family-benchmark_rich_background.owl"
# Available OWL Reasoners: 'HermiT', 'Pellet', 'JFact', 'Openllet'
sync_reasoner = SyncReasoner(ontology = ontology_path, reasoner="Pellet")
onto = OntologyManager().load_ontology(ontology_path)
# Iterate over defined owl Classes in the signature
for i in onto.classes_in_signature():
# Performing type inference with Pellet
instances=sync_reasoner.instances(i,direct=False)
print(f"Class:{i}\t Num instances:{len(instances)}")
stopJVM()
Click me!
An Ontology can be enriched by inferring many different axioms.
from owlapy.owl_reasoner import SyncReasoner
from owlapy.static_funcs import stopJVM
sync_reasoner = SyncReasoner(ontology="KGs/Family/family-benchmark_rich_background.owl", reasoner="Pellet")
# Infer missing class assertions
sync_reasoner.infer_axioms_and_save(output_path="KGs/Family/inferred_family-benchmark_rich_background.ttl",
output_format="ttl",
inference_types=[
"InferredClassAssertionAxiomGenerator",
"InferredEquivalentClassAxiomGenerator",
"InferredDisjointClassesAxiomGenerator",
"InferredSubClassAxiomGenerator",
"InferredInverseObjectPropertiesAxiomGenerator",
"InferredEquivalentClassAxiomGenerator"])
stopJVM()
Check also the examples and tests folders.
Currently, we are working on our manuscript describing our framework.