smeoni commited on
Commit
e957396
·
1 Parent(s): 08e8e81

Upload e3c.py

Browse files
Files changed (1) hide show
  1. e3c.py +12 -5
e3c.py CHANGED
@@ -30,10 +30,10 @@ _URL = "https://github.com/hltfbk/E3C-Corpus/archive/refs/tags/v2.0.0.zip"
30
 
31
 
32
  class E3CConfig(datasets.BuilderConfig):
33
- """BuilderConfig for NCBIDisease"""
34
 
35
  def __init__(self, **kwargs):
36
- """BuilderConfig for NCBIDisease.
37
  Args:
38
  **kwargs: keyword arguments forwarded to super.
39
  """
@@ -46,7 +46,7 @@ class E3C(datasets.GeneratorBasedBuilder):
46
  E3CConfig(
47
  name="e3c",
48
  version=VERSION,
49
- description="this is the split of the layer 1 for English of E3C dataset",
50
  ),
51
  ]
52
 
@@ -54,7 +54,9 @@ class E3C(datasets.GeneratorBasedBuilder):
54
  """This method specifies the DatasetInfo which contains information and typings."""
55
  features = datasets.Features(
56
  {
 
57
  "tokens": datasets.Sequence(datasets.Value("string")),
 
58
  "ner_tags": datasets.Sequence(
59
  datasets.features.ClassLabel(
60
  names=[
@@ -263,8 +265,8 @@ class E3C(datasets.GeneratorBasedBuilder):
263
  Args:
264
  filepath: The path to the folder containing the files to parse.
265
  Yields:
266
- The unique id of an example and the example itself containing tokens and ner_tags in
267
- IOB format.
268
  """
269
  guid = 0
270
  for content in self.get_parsed_data(filepath):
@@ -275,6 +277,9 @@ class E3C(datasets.GeneratorBasedBuilder):
275
  content["TOKENS"],
276
  )
277
  )
 
 
 
278
  labels = ["O"] * len(filtered_tokens)
279
  for entity_type in [
280
  "CLINENTITY",
@@ -302,7 +307,9 @@ class E3C(datasets.GeneratorBasedBuilder):
302
  else:
303
  labels[idx_token] = f"{entity_type}"
304
  yield guid, {
 
305
  "tokens": list(map(lambda tokens: tokens[2], filtered_tokens)),
306
  "ner_tags": labels,
 
307
  }
308
  guid += 1
 
30
 
31
 
32
  class E3CConfig(datasets.BuilderConfig):
33
+ """BuilderConfig for E3C."""
34
 
35
  def __init__(self, **kwargs):
36
+ """BuilderConfig for E3C.
37
  Args:
38
  **kwargs: keyword arguments forwarded to super.
39
  """
 
46
  E3CConfig(
47
  name="e3c",
48
  version=VERSION,
49
+ description="this is an implementation of the E3C dataset",
50
  ),
51
  ]
52
 
 
54
  """This method specifies the DatasetInfo which contains information and typings."""
55
  features = datasets.Features(
56
  {
57
+ "text": datasets.Value("string"),
58
  "tokens": datasets.Sequence(datasets.Value("string")),
59
+ "tokens_offsets": datasets.Sequence(datasets.Sequence(datasets.Value("int32"))),
60
  "ner_tags": datasets.Sequence(
61
  datasets.features.ClassLabel(
62
  names=[
 
265
  Args:
266
  filepath: The path to the folder containing the files to parse.
267
  Yields:
268
+ an example containing four fields: the text, the annotations, the tokens offsets and
269
+ the sentences.
270
  """
271
  guid = 0
272
  for content in self.get_parsed_data(filepath):
 
277
  content["TOKENS"],
278
  )
279
  )
280
+ tokens_offsets = [
281
+ [token[0] - sentence[0], token[1] - sentence[0]] for token in filtered_tokens
282
+ ]
283
  labels = ["O"] * len(filtered_tokens)
284
  for entity_type in [
285
  "CLINENTITY",
 
307
  else:
308
  labels[idx_token] = f"{entity_type}"
309
  yield guid, {
310
+ "text": sentence[-1],
311
  "tokens": list(map(lambda tokens: tokens[2], filtered_tokens)),
312
  "ner_tags": labels,
313
+ "tokens_offsets": tokens_offsets,
314
  }
315
  guid += 1