Update the csv order

ConceptLearner/DiscriminativeExplainer.py

@@ -21,24 +21,6 @@ from typing import Final
 from ontolearn.abstracts import AbstractFitness
 from ontolearn.ea_utils import Tree
 
-class LinearPressureFitness(AbstractFitness):
-    """Linear parametric parsimony pressure."""
-
-    __slots__ = 'gain', 'penalty'
-
-    name: Final = 'Linear_Pressure_Fitness'
-
-    def __init__(self, gain: float = 2048.0, penalty: float = 1.0):
-        self.gain = gain
-        self.penalty = penalty
-
-    def apply(self, individual: Tree):
-        quality = individual.quality.values[0]
-        fitness = self.gain*quality - self.penalty*len(individual)
-                
-        # print(self.gain, quality, self.gain*quality, len(individual))
-        
-        individual.fitness.values = (round(fitness, 5),)
 
 class DiscriminativeExplainer:
     """ An abstract class which represent an explainer. An explainer should be able to use a label to generate a
@@ -86,7 +68,7 @@ class DiscriminativeExplainer:
             return OWLObjectMinCardinality(ce.get_cardinality(), ce.get_property(), ce2)
         return ce
 
-    def __init__(self, gnn, data: Union[HeteroData, Data], namespace = "http://example.org/", owl_graph_path = "./owlGraphs/example.owl", generate_new_owl_file: bool = False, create_nominals: bool = False,  add_edge_counts: bool = False, create_data_properties_as_boolean: bool = False, full_edge_name: bool = False, ignore_nodes: bool = False, high_level_concepts: dict = None, create_high_level_concepts_as_boolean: bool = False,) -> None:
+    def __init__(self, gnn, data: Union[HeteroData, Data], namespace = "http://example.org/", owl_graph_path = "./owlGraphs/example.owl", generate_new_owl_file: bool = False, create_nominals: bool = False,  add_edge_counts: bool = False, create_data_properties_as_boolean: bool = False, full_edge_name: bool = False, ignore_nodes: bool = False, high_level_concepts: dict = None, create_high_level_concepts_as_boolean: bool = False, add_false_values=False) -> None:
         """Initializes the explainer based on the given GNN and the Dataset. After the initialization the object should
         be able to produce explanations of single labels.
 
@@ -115,7 +97,7 @@ class DiscriminativeExplainer:
         if generate_new_owl_file and os.path.isfile(self.owl_graph_path):
             os.remove(self.owl_graph_path)
         if not os.path.isfile(self.owl_graph_path):
-            self.owlGraph = convert_to_owl(data=self.data, namespace=self.namespace, owlGraphPath=self.owl_graph_path, high_level_concepts=self.high_level_concepts, create_nominals=create_nominals, add_edge_counts=add_edge_counts, create_data_properties_as_boolean = create_data_properties_as_boolean, full_edge_name=full_edge_name, ignore_nodes=self.ignore_nodes, create_high_level_concepts_as_boolean=create_high_level_concepts_as_boolean)
+            self.owlGraph = convert_to_owl(data=self.data, namespace=self.namespace, owlGraphPath=self.owl_graph_path, high_level_concepts=self.high_level_concepts, create_nominals=create_nominals, add_edge_counts=add_edge_counts, create_data_properties_as_boolean = create_data_properties_as_boolean, full_edge_name=full_edge_name, ignore_nodes=self.ignore_nodes, create_high_level_concepts_as_boolean=create_high_level_concepts_as_boolean, add_false_values=add_false_values)
             self.owlGraph.buildGraph()
         self.knowledge_base = KnowledgeBase(path=self.owl_graph_path)
 
@@ -126,8 +108,7 @@ class DiscriminativeExplainer:
             debug: Optional[bool] = False,
             max_runtime: Optional[int] = 60,
             num_generations: Optional[int] = 600,
-            quality_func: Optional[AbstractScorer] = None,
-            length_penalty: Optional[int] = 1.0) -> OWLClassExpression:
+            quality_func: Optional[AbstractScorer] = None) -> OWLClassExpression:
         """Explains based on the GNN a given label. The explanation is in the form of a Class Expression.
 
         Args:
@@ -144,7 +125,7 @@ class DiscriminativeExplainer:
 
         if quality_func is None:
             quality_func = F1()
-        self.model = EvoLearner(knowledge_base=self.knowledge_base, use_data_properties=use_data_properties, max_runtime=max_runtime, num_generations=num_generations, quality_func=quality_func, population_size=1000, fitness_func=LinearPressureFitness(penalty=length_penalty))
+        self.model = EvoLearner(knowledge_base=self.knowledge_base, use_data_properties=use_data_properties, max_runtime=max_runtime, num_generations=num_generations, quality_func=quality_func, population_size=1000)
         positive_examples = []
         negative_examples = []
         

text_based_datasets_experiment.py

@@ -69,8 +69,17 @@ def append_to_csv_file(results, results_dir, dataset_key, num_groups, create_hig
     filename = os.path.join(results_dir, f"{dataset_key}_results.csv")
     
     with open(filename, mode='a', newline='', encoding='utf-8') as csvfile:
-        fieldnames = ['Dataset', 'Number of Groups', 'Create High Level Concepts As Boolean', 'Label Name', 
-                      'Hypothesis', 'Accuracy', 'Recall', 'Precision', 'F1 Score', 'Length', 'Explain Time']
+        fieldnames = ['Dataset', 
+                      'Number of Groups', 
+                      'F1 Score', 
+                      'Length', 
+                      'Create High Level Concepts As Boolean', 
+                      'Label Name', 
+                      'Hypothesis', 
+                      'Accuracy', 
+                      'Recall', 
+                      'Precision',
+                      'Explain Time']
         writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
         
         if write_header:
@@ -80,14 +89,14 @@ def append_to_csv_file(results, results_dir, dataset_key, num_groups, create_hig
             writer.writerow({
                 'Dataset': dataset_key,
                 'Number of Groups': num_groups,
+                'F1 Score': data['evaluation'].get('F1', 'N/A'),
+                'Length': data['length'],
                 'Create High Level Concepts As Boolean': create_high_level_concepts_as_boolean,
                 'Label Name': data['label_name'],
                 'Hypothesis': data['hypothesis'],
                 'Accuracy': data['evaluation'].get('Accuracy', 'N/A'),
                 'Recall': data['evaluation'].get('Recall', 'N/A'),
                 'Precision': data['evaluation'].get('Precision', 'N/A'),
-                'F1 Score': data['evaluation'].get('F1', 'N/A'),
-                'Length': data['length'],
                 'Explain Time': data['explain_time']
             })
 
@@ -157,9 +166,16 @@ def summarize_aggregated_results(aggregated_results, results_dir, dataset_name):
 
     with open(summary_filename, mode="w", newline="", encoding="utf-8") as csvfile:
         fieldnames = [
-            "Label Name", "Number of Groups", "Best Hypothesis", "Best F1 Score",
-            "Average F1 Score", "Best Accuracy", "Average Accuracy", "Length at Best F1",
-            "Average Length", "Average Explain Time"
+            "Label Name",
+            "Number of Groups",
+            "Best F1 Score",
+            "Average F1 Score",
+            "Best Accuracy",
+            "Average Accuracy",
+            "Length at Best F1",
+            "Average Length",
+            "Best Hypothesis",
+            "Average Explain Time"
         ]
         writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
         writer.writeheader()
@@ -173,13 +189,13 @@ def summarize_aggregated_results(aggregated_results, results_dir, dataset_name):
             writer.writerow({
                 "Label Name": data["label_name"],
                 "Number of Groups": num_groups,
-                "Best Hypothesis": data["best_hypothesis"],
                 "Best F1 Score": data["best_F1"],
                 "Average F1 Score": avg_f1,
                 "Best Accuracy": data["best_accuracy"],
                 "Average Accuracy": avg_accuracy,
                 "Length at Best F1": data["length_at_best_f1"],
                 "Average Length": avg_length,
+                "Best Hypothesis": data["best_hypothesis"],
                 "Average Explain Time": avg_time
             })