OntoLearner 1.4.10__py3-none-any.whl → 1.5.0__py3-none-any.whl

ontolearner/ontology/biology.py

@@ -70,15 +70,14 @@ class GO(BaseOntology):
     format = "OWL"
     download_url = "https://geneontology.org/docs/download-ontology/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle GO-specific blank nodes."""
         # Check the general patterns from the parent class
         # GO-specific patterns
         if label.startswith('GO_'):
             return True
 
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/ontology/chemistry.py

@@ -57,15 +57,15 @@ class ChEBI(BaseOntology):
     format = "OWL"
     download_url = "https://www.ebi.ac.uk/chebi/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle ChEBI-specific blank nodes."""
         # ChEBI-specific patterns
         if re.match(r'^CHEBI_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -89,15 +89,15 @@ class CHEMINF(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/CHEMINF"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle CHEMINF-specific blank nodes."""
         # ChEBI-specific patterns
         if re.match(r'^CHEMINF_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -145,15 +145,15 @@ class ChMO(BaseOntology):
     format = "OWL"
     download_url = "https://github.com/rsc-ontologies/rsc-cmo"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle ChMO-specific blank nodes."""
         # ChEBI-specific patterns
         if re.match(r'^CHMO_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -190,15 +190,15 @@ class MassSpectrometry(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/MS"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle MassSpectrometry-specific blank nodes."""
         # MassSpectrometry-specific patterns
         if re.match(r'^PEFF_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -220,8 +220,7 @@ class MOP(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/MOP"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle MOP-specific blank nodes."""
         # MOP-specific patterns
         if re.match(r'^MOP_[0-9]+$', label):
@@ -234,7 +233,7 @@ class MOP(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -357,8 +356,7 @@ class RXNO(BaseOntology):
     format = "OWL"
     download_url = "https://github.com/rsc-ontologies/rxno"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle RXNO-specific blank nodes."""
         # RXNO-specific patterns
         if re.match(r'^RXNO_[0-9]+$', label):
@@ -368,7 +366,7 @@ class RXNO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/ontology/ecology_environment.py

@@ -35,15 +35,14 @@ class ENVO(BaseOntology):
     format = "OWL"
     download_url = "https://obofoundry.org/ontology/envo.html"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle ENVO-specific blank nodes."""
         # ENVO-specific patterns
         if re.match(r'^PATO_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/ontology/general.py

@@ -53,15 +53,14 @@ class DBpedia(BaseOntology):
     format = "OWL"
     download_url = "https://wiki.dbpedia.org/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle DBpedia/Wikidata-specific blank nodes."""
         # DBpedia/Wikidata-specific patterns
         if re.match(r'^Q[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -226,15 +225,14 @@ class UMBEL(BaseOntology):
     format = "n3"
     download_url = "https://github.com/structureddynamics/UMBEL/tree/master/Ontology"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle UMBEL-specific blank nodes."""
         # UMBEL-specific patterns
         if re.match(r'^f5295f96ac3e649dcb1740b0d93d3e6c2b[0-9a-f]+$', label):  # Long hexadecimal identifiers
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/ontology/material_science_engineering.py

@@ -74,15 +74,14 @@ class Atomistic(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/domain-atomistic"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle Atomistic-specific blank nodes."""
         # EMMO-specific patterns (UUID format) in Atomistic
         if re.match(r'^EMMO_[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -106,8 +105,8 @@ class BattINFO(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/BIG-MAP/BattINFO"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle BattINFO-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -123,7 +122,7 @@ class BattINFO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -168,8 +167,7 @@ class BVCO(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/Battery-Value-Chain-Ontology/ontology"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle BVCO-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -184,7 +182,7 @@ class BVCO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -277,8 +275,7 @@ class CHAMEO(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/domain-characterisation-methodology"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle CHAMEO-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -287,7 +284,7 @@ class CHAMEO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -400,8 +397,7 @@ class EMMOCrystallography(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/domain-crystallography"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle EMMOCrystallography-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -410,7 +406,7 @@ class EMMOCrystallography(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -451,8 +447,7 @@ class GPO(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/General-Process-Ontology/ontology"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle GPO-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -463,7 +458,7 @@ class GPO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -697,8 +692,7 @@ class MechanicalTesting(BaseOntology):
     format = "OWL"
     download_url = "https://github.com/emmo-repo/domain-mechanical-testing"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle MechanicalTesting-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -707,7 +701,7 @@ class MechanicalTesting(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -731,8 +725,7 @@ class MicroStructures(BaseOntology):
     format = "OWL"
     download_url = "https://github.com/jesper-friis/emmo-microstructure"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle MicroStructures-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -741,7 +734,7 @@ class MicroStructures(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -868,8 +861,7 @@ class OIECharacterisation(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/OIE-Ontologies/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle OIECharacterisation-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -878,7 +870,7 @@ class OIECharacterisation(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -901,8 +893,7 @@ class OIEManufacturing(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/OIE-Ontologies/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle OIEManufacturing-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -911,7 +902,7 @@ class OIEManufacturing(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -934,8 +925,7 @@ class OIEMaterials(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/OIE-Ontologies/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle OIEMaterials-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -944,7 +934,7 @@ class OIEMaterials(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -967,8 +957,7 @@ class OIEModels(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/OIE-Ontologies/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle OIEModels-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -977,7 +966,7 @@ class OIEModels(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -999,8 +988,7 @@ class OIESoftware(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/OIE-Ontologies/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle OIESoftware-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -1009,7 +997,7 @@ class OIESoftware(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -1067,8 +1055,7 @@ class Photovoltaics(BaseOntology):
     format = "TTL"
     download_url = "https://github.com/emmo-repo/domain-photovoltaics"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle Photovoltaics-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -1077,7 +1064,7 @@ class Photovoltaics(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -1219,8 +1206,8 @@ class VIMMP(BaseOntology):
     format = "OWL"
     download_url = "https://matportal.org/ontologies/VIMMP_ONTOLOGIES"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle VIMMP-specific blank nodes."""
         # UUID pattern for various prefixes
         uuid_pattern = r'[0-9a-f]{8}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{4}_[0-9a-f]{12}$'
@@ -1232,7 +1219,39 @@ class VIMMP(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
+
+
+class MDSOnto(BaseOntology):
+    """
+    MDS-Onto is a domain (low) level ontology that describes terms in Materials Data Science. It is divided into six
+    large modules: BuiltEnv, Exposure, Chemistry, Manufacture, Characterization, and Geospatial. Under each module,
+    there are multiple sub-modules such as FTIR, AFM, Chem-Rxn, PV-Module, Accelerated Exposure, etc.
+    """
+    ontology_id = "MDSOnto"
+    ontology_full_name = "The Modular Ontology for Materials and Data Science (MDS-Onto)"
+    domain = "Materials Science and Engineering"
+    category = "Materials Data Science"
+    version = "0.3.1.16"
+    last_updated = "2026-02-03"
+    creator = "SDLE Research Center"
+    license = "CC BY-SA 4.0"
+    format = "OWL"
+    download_url = "https://cwrusdle.bitbucket.io/files/MDS_Onto/index-en.html"
+
+    def _is_anonymous_id(self, label: str) -> bool:
+        """Override to handle MDSOnto-specific and ontology-generated blank nodes."""
+        # Treat IDs starting with 'ont0' as anonymous (e.g. ont00000562)
+        if re.match(r'^ont0', label):
+            return True
+        if super()._is_anonymous_id(label):
+            return True
+        return False
+
+
+    # def contains_imports(self) -> bool:
+    #     """Hook: Check if the ontology contains imports."""
+    #     return True  # Set to True if your ontology imports other ontologies

ontolearner/ontology/medicine.py

@@ -33,14 +33,13 @@ class BTO(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/BTO"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle VIMMP-specific blank nodes."""
         if re.match(r'^BTO_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/ontology/scholarly_knowledge.py

@@ -181,8 +181,7 @@ class DUO(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/DUO/"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle DUO-specific blank nodes."""
         if re.match(r'^APOLLO_SV_[0-9]+$', label):
             return True
@@ -191,7 +190,7 @@ class DUO(BaseOntology):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -327,14 +326,13 @@ class Metadata4Ing(BaseOntology):
     format = "TTL"
     download_url = "https://git.rwth-aachen.de/nfdi4ing/metadata4ing/metadata4ing"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle Metadata4Ing-specific blank nodes."""
         if re.match(r'^\d{4}-\d{4}-\d{4}-\d{4}$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False
@@ -517,14 +515,13 @@ class SWO(BaseOntology):
     format = "OWL"
     download_url = "https://terminology.tib.eu/ts/ontologies/SWO"
 
-    @staticmethod
-    def _is_anonymous_id(label: str) -> bool:
+    def _is_anonymous_id(self, label: str) -> bool:
         """Override to handle SWO-specific blank nodes."""
         if re.match(r'^SWO_[0-9]+$', label):
             return True
 
         # Check the general patterns from the parent class
-        if BaseOntology._is_anonymous_id(label):
+        if super()._is_anonymous_id(label):
             return True
 
         return False

ontolearner/processor.py

@@ -850,13 +850,13 @@ If you find our work helpful, feel free to give us a cite.
 
             # Commit and push ontology repository
             repo.git_add(auto_lfs_track=True)
-            commit_message = f":sparkles: Added {self.ontology_id} ontology!"
+            commit_message = f"✨ Added {self.processed_ontology['ontology_id']} ontology!"
             repo.git_commit(commit_message)
             repo.git_push()
 
             # Commit and push metrics repository
             metrics_repo.git_add(auto_lfs_track=True)
-            metrics_commit_message = f":memo: Update metrics for {self.ontology_id}"
+            metrics_commit_message = f"📝 Update metrics for {self.processed_ontology['ontology_id']}"
             metrics_repo.git_commit(metrics_commit_message)
             metrics_repo.git_push()
 
@@ -864,7 +864,7 @@ If you find our work helpful, feel free to give us a cite.
                 "status": "success",
                 "repository": repo_id,
                 "metrics_repository": metrics_repo_id,
-                "ontology_id": self.ontology_id,
+                "ontology_id": self.processed_ontology['ontology_id'],
                 "url": f"https://huggingface.co/datasets/{repo_id}",
                 "metrics_url": f"https://huggingface.co/spaces/{metrics_repo_id}"
             }

ontolearner/text2onto/splitter.py

@@ -200,10 +200,73 @@ class SyntheticDataSplitter:
 
         return terms_splits, types_splits, docs_split, types2docs_splits
 
-    def split(self, train: float = 0.8, val: float = 0.1, test: float = 0.1):
-        split_targets, split_docs_targets = self.set_train_val_test_sizes(train_percentage=train,
-                                                                          val_percentage=val,
-                                                                          test_percentage=test)
+    def split_fine_grained(self, doc_ids):
+        """
+        Build a single split bundle containing only:
+        - docs
+        - terms
+        - types
+        - terms2docs
+        - terms2types
+        """
+        # normalize to string ids (constructor uses str(row.id))
+        doc_ids = {str(d) for d in (doc_ids or [])}
+
+        # docs + collect terms/types from docs
+        docs = []
+        terms_set = set()
+        types_set = set()
+
+        for doc_id in doc_ids:
+            doc = self.doc_id_to_doc[doc_id]
+            docs.append({"id": str(doc.id), "title": doc.title, "text": doc.text})
+
+            terms_set.update(self.doc_id_to_terms[doc_id])
+            types_set.update(self.doc_id_to_types[doc_id])
+
+        terms = sorted(terms_set)
+        types = sorted(types_set)
+
+        # terms2docs: use the constructor-built mapping and restrict to this split's doc_ids
+        terms2docs = {
+            term: sorted(list(self.term_to_doc_id.get(term, set()) & doc_ids))
+            for term in terms
+        }
+
+        # terms2types: ontology lookup (term -> parent types)
+        terms2types = {term: self.child_to_parent.get(term, []) for term in terms}
+
+        return {
+            "documents": docs,
+            "terms": terms,
+            "types": types,
+            "terms2docs": terms2docs,
+            "terms2types": terms2types,
+        }
+
+    def train_test_val_split(self, train: float = 0.8, val: float = 0.1, test: float = 0.1):
+        """
+        Returns:
+            train_split, val_split, test_split
+
+        Each split is a dict with keys:
+        - "docs"
+        - "terms"
+        - "types"
+        - "terms2docs"
+        - "terms2types"
+        """
+        # compute which docs go to which split
+        split_targets, split_docs_targets = self.set_train_val_test_sizes(
+            train_percentage=train,
+            val_percentage=val,
+            test_percentage=test,
+        )
         split_docs = self.create_train_val_test_splits(split_targets, split_docs_targets)
-        terms, types, docs, types2docs = self.generate_split_artefacts(split_docs)
-        return terms, types, docs, types2docs
+        # split_docs: {"train": set(doc_ids), "val": set(doc_ids), "test": set(doc_ids)}
+
+        train_split = self.split_fine_grained(split_docs.get("train", set()))
+        val_split   = self.split_fine_grained(split_docs.get("val", set()))
+        test_split  = self.split_fine_grained(split_docs.get("test", set()))
+
+        return train_split, val_split, test_split