WIP: Cardinality restriction Experimentations

ontolearn/triple_store.py

@@ -1018,48 +1018,8 @@ def instances(self, expression: OWLClassExpression, named_individuals: bool = Fa
 
             yield from results
 
-        # elif isinstance(expression, OWLObjectMaxCardinality):
-
-        #     property_expression = expression.get_property()
-        #     filler_expression = expression.get_filler()
-
-        #     # (1) Find individuals that are likely y \in C^I, i.e. φ(y, type, C) 
-
-        #     filler_individuals = {i for i in self.instances(filler_expression)}
-
-        #     # 2) set max cardinality
-        #     max_cardinality = expression.get_cardinality()
-
-        #     # (3) Iterative (1) and return entities whose predicted score satisfies the condition.
-        #     results = set()
-        #     for individual in filler_individuals:# self.owl_individuals:
-
-        #         if isinstance(property_expression, OWLObjectInverseOf):
-
-        #             inverse_property = property_expression.get_inverse_property()
-        #             scores = self.neural_link_predictor.predict(h=[individual.str],
-        #                                                         r=[inverse_property.str],
-        #                                                         within=None,
-        #                                                         logits=False).tolist()
-        #         else:
-        #             scores = self.neural_link_predictor.predict(r=[property_expression.str],
-        #                                                         t=[individual.str],
-        #                                                         within=None,
-        #                                                         logits=False).tolist()
-
-        #         candidate_iris = [iri for iri, score in zip(self.neural_link_predictor.entity_to_idx.keys(), scores) if score >= self.gamma_for_nc]
-
-        #         count = len(candidate_iris)
-
-        #         if count <= max_cardinality:
-        #             for iri in candidate_iris:
-        #                 try:
-        #                     results.add(OWLNamedIndividual(iri))
-        #                 except:
-        #                     print(f"Could not convert to OWLNamedIndividual {iri}")
-
-        #     yield from results
-
+
+
 
         elif isinstance(expression, OWLObjectMaxCardinality):
 
@@ -1071,8 +1031,6 @@ def instances(self, expression: OWLClassExpression, named_individuals: bool = Fa
             # 1) define max cardinality
             max_cardinality = expression.get_cardinality()
 
-            filler_individuals = {i for i in self.instances(filler_expression)}
-
             # (2) Iterative (1) and return entities whose predicted score satisfies the condition.
             results = set()
             for individual in self.owl_individuals:
@@ -1097,8 +1055,12 @@ def instances(self, expression: OWLClassExpression, named_individuals: bool = Fa
                 count = sum(1 for idx, score in enumerate(scores) if score >= self.gamma_for_nc and
                             self.is_instance(self.neural_link_predictor.idx_to_entity[idx], filler_expression))
 
-                if count <= max_cardinality:
-                    results.add(individual)
+                if max_cardinality == 0:
+                    raise NotImplementedError("Handling for max cardinality of 0 is not implemented.")
+
+                else:
+                    if count <= max_cardinality:
+                        results.add(individual)
 
             yield from results
 

tests/test_concept_EBR.py

@@ -215,8 +215,8 @@ def get_reasoner_instances(reasoner: NeuralReasoner, class_expression: OWLClassE
 
 print(len(ground_truth_1))
 print(len(ground_truth_2))
-print(len(reasoner_instances_2))
 print(len(reasoner_instances_1))
+print(len(reasoner_instances_2))
 
 assert reasoner_instances_1 == reasoner_instances_2, "Reasoner fails"
 print("test 6 done")