Update MANTRAGSC.py

Fixed missing labels for other languages and updated labels order to prepare for parquet conversion.

MANTRAGSC.py

@@ -25,7 +25,6 @@ import xmltodict
 import numpy as np
 
 import datasets
-from filelock import FileLock
 
 _CITATION = """\
 @article{10.1093/jamia/ocv037,
@@ -86,6 +85,7 @@ _DATASET_TYPES = {
     "patents": "Patent",
 }
 
+
 @dataclass
 class DrBenchmarkConfig(datasets.BuilderConfig):
     name: str = None
@@ -94,6 +94,7 @@ class DrBenchmarkConfig(datasets.BuilderConfig):
     schema: str = None
     subset_id: str = None
 
+
 class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
     SOURCE_VERSION = datasets.Version("1.0.0")
@@ -102,12 +103,13 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
     for language, dataset_type in product(_LANGUAGES_2, _DATASET_TYPES):
 
-        if dataset_type == "patents" and language in ["nl", "es"]:
+        name = f"{language}_{dataset_type}"
+        if name in ['nl_patents', 'es_patents', 'en_medline']:
             continue
 
         BUILDER_CONFIGS.append(
             DrBenchmarkConfig(
-                name=f"{language}_{dataset_type}",
+                name=name,
                 version=SOURCE_VERSION,
                 description=f"Mantra GSC {_LANGUAGES_2[language]} {_DATASET_TYPES[dataset_type]} source schema",
                 schema="source",
@@ -118,13 +120,34 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
     DEFAULT_CONFIG_NAME = "fr_medline"
 
     def _info(self):
-
-        if self.config.name.find("emea") != -1:
-            names = ['B-ANAT', 'I-ANAT', 'I-PHEN', 'B-PROC', 'I-CHEM', 'I-PHYS', 'B-DEVI', 'O', 'B-PHYS', 'I-DEVI', 'B-OBJC', 'I-DISO', 'B-PHEN', 'I-LIVB', 'B-DISO', 'B-LIVB', 'B-CHEM', 'I-PROC']
-        elif self.config.name.find("medline") != -1:
-            names = ['B-ANAT', 'I-ANAT', 'B-PROC', 'I-CHEM', 'I-PHYS', 'B-GEOG', 'B-DEVI', 'O', 'B-PHYS', 'I-LIVB', 'B-OBJC', 'I-DISO', 'I-DEVI', 'B-PHEN', 'B-DISO', 'B-LIVB', 'B-CHEM', 'I-PROC']
-        elif self.config.name.find("patents") != -1:
-            names = ['B-ANAT', 'I-ANAT', 'B-PROC', 'I-CHEM', 'I-PHYS', 'B-DEVI', 'O', 'I-LIVB', 'B-OBJC', 'I-DISO', 'B-PHEN', 'I-PROC', 'B-DISO', 'I-DEVI', 'B-LIVB', 'B-CHEM', 'B-PHYS']
+        # Label definition for each task
+        # Goals:
+        # - Tasks must not have extra labels (not present in their corpus)
+        # - Labels should have (roughly) the same index
+
+        # Labels common to every task (ordered by name and B-I)
+        common_names = ['O', 'B-ANAT', 'B-CHEM', 'I-CHEM', 'B-DEVI', 'B-DISO', 'I-DISO', 'B-LIVB', 'I-LIVB', 'B-OBJC', 'B-PHEN', 'B-PHYS', 'I-PHYS', 'B-PROC', 'I-PROC']
+        # Adding labels not common to every task (in an order that maximises labels having the same index accross tasks)
+        names = common_names + ["I-ANAT", "I-DEVI", "B-GEOG", "I-PHEN", "I-OBJC"]
+        unused_name_map = {
+            'de_emea': {'B-GEOG', 'I-OBJC'},
+            'en_emea': {'B-GEOG', 'I-OBJC'},
+            'es_emea': {'B-GEOG', 'I-OBJC'},
+            'fr_emea': {'B-GEOG', 'I-OBJC'},
+            'nl_emea': {'B-GEOG', 'I-OBJC'},
+
+            'de_medline': {'I-DEVI', 'I-PHEN'},
+            'es_medline': {'I-DEVI', 'I-OBJC'},
+            'fr_medline': {'I-OBJC', 'I-PHEN'},
+            'nl_medline': {'I-DEVI'},
+
+            'fr_patents': {'B-GEOG', 'I-OBJC', 'I-PHEN'},
+            'de_patents': {'B-GEOG', 'I-OBJC', 'I-PHEN', 'I-ANAT', 'I-DEVI'},
+            'en_patents': {'B-GEOG', 'I-OBJC', 'I-PHEN'}
+        }
+        names = [n for n in names if n not in unused_name_map.get(self.config.name, {})]
+
+        print(self.config.name)
 
         features = datasets.Features(
             {
@@ -132,7 +155,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
                 "tokens": [datasets.Value("string")],
                 "ner_tags": datasets.Sequence(
                     datasets.features.ClassLabel(
-                        names = names,
+                        names=names,
                     )
                 ),
             }
@@ -150,10 +173,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
         language, dataset_type = self.config.name.split("_")
 
-        # Fixes concurrency issues when extracting files after download has ended
-        # cf. https://github.com/huggingface/datasets/issues/4661#issuecomment-2792885416
-        with FileLock(Path(datasets.config.HF_CACHE_HOME) / "tmp_MANTRAGSC.lock"):
-            data_dir = dl_manager.download_and_extract(_URL)
+        data_dir = dl_manager.download_and_extract(_URL)
         data_dir = Path(data_dir) / "GSC-v1.1" / f"{_DATASET_TYPES[dataset_type]}_GSC_{language}_man.xml"
 
         return [
@@ -189,7 +209,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
         for d in doc["Corpus"]["document"]:
 
-            if type(d["unit"]) != type(list()):
+            if not isinstance(d["unit"], list):
                 d["unit"] = [d["unit"]]
 
             for u in d["unit"]:
@@ -198,7 +218,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
                 if "e" in u.keys():
 
-                    if type(u["e"]) != type(list()):
+                    if not isinstance(u["e"], list):
                         u["e"] = [u["e"]]
 
                     tags = [{
@@ -214,7 +234,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
                 tokens = []
                 for i, t in enumerate(_tokens):
 
-                    concat = " ".join(_tokens[0:i+1])
+                    concat = " ".join(_tokens[0:i + 1])
 
                     offset_start = len(concat) - len(t)
                     offset_end = len(concat)
@@ -233,8 +253,8 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
 
                     for idx, token in enumerate(tokens):
 
-                        rtok = range(token["offset_start"], token["offset_end"]+1)
-                        rtag = range(tag["offset_start"], tag["offset_end"]+1)
+                        rtok = range(token["offset_start"], token["offset_end"] + 1)
+                        rtag = range(tag["offset_start"], tag["offset_end"] + 1)
 
                         # Check if the ranges are overlapping
                         if bool(set(rtok) & set(rtag)):
@@ -273,7 +293,7 @@ class MANTRAGSC(datasets.GeneratorBasedBuilder):
         random.shuffle(ids)
         random.shuffle(ids)
 
-        train, validation, test = np.split(ids, [int(len(ids)*0.70), int(len(ids)*0.80)])
+        train, validation, test = np.split(ids, [int(len(ids) * 0.70), int(len(ids) * 0.80)])
 
         if split == "train":
             allowed_ids = list(train)