subcorpus exportation

config/options_menu.yaml

@@ -16,7 +16,7 @@ root:
     - PU_learning
     - get_feedback_options
     - update_model
-    - import_export_annotations
+    - import_export_data
     - inference
 
 # In the following, each menu option is described.
@@ -123,11 +123,12 @@ update_model:
     - reevaluate_model
     - performance_metrics_PN
 
-import_export_annotations:
-  title: Import / export annotations
+import_export_data:
+  title: Import / export data
   options:
     - import_annotations
     - export_annotations
+    - export_subcorpus
 
 inference:
   title: Apply inference over precomputed embeddings (if available, only)
@@ -227,3 +228,7 @@ import_annotations:
 
 export_annotations:
   title: "Export annotations (warning: deletes older annotation files)"
+
+export_subcorpus:
+  title: "Export subcorpus"
+

src/task_manager.py

@@ -1418,6 +1418,29 @@ def export_annotations(self, domain_name: str):
 
         return
 
+    def export_subcorpus(self):
+        """
+        Exports the list of IDs corresponding to documents from the positive
+        class
+        """
+
+        if self.df_dataset is None:
+            logging.warning("-- No model is loaded. "
+                            "You must load or create a set of labels first")
+            return
+
+        subcorpus = self.df_dataset[self.df_dataset.prediction == 1]
+
+        # Save ids only
+        path2parquet = (
+            self.path2output / f'subcorpus_{self.class_name}.parquet')
+        path2csv = (
+            self.path2output / f'subcorpus_{self.class_name}.csv')
+        subcorpus[['id']].to_parquet(path2parquet)
+        subcorpus[['id']].to_csv(path2csv)
+
+        return
+
 
 class TaskManagerCMD(TaskManager):
     """