Datasets:

Modalities:
Text
Languages:
Spanish
Libraries:
Datasets
License:
ccasimiro commited on
Commit
db0a277
1 Parent(s): 11f5c7c

upload dataset

Browse files
Files changed (6) hide show
  1. .gitattributes +3 -0
  2. README.md +132 -0
  3. dev.conll +3 -0
  4. pharmaconer.py +133 -0
  5. test.conll +3 -0
  6. train.conll +3 -0
.gitattributes CHANGED
@@ -25,3 +25,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
28
+ dev.conll filter=lfs diff=lfs merge=lfs -text
29
+ test.conll filter=lfs diff=lfs merge=lfs -text
30
+ train.conll filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - expert-generated
4
+ languages:
5
+ - es
6
+ multilinguality:
7
+ - monolingual
8
+ task_categories:
9
+ - text-classification
10
+ - multi-label-text-classification
11
+ task_ids:
12
+ - named-entity-recognition
13
+ ---
14
+
15
+ # PharmaCoNER Corpus
16
+
17
+ ## BibTeX citation
18
+ If you use these resources in your work, please cite the following paper:
19
+
20
+ ```bibtex
21
+ TO DO
22
+ ```
23
+
24
+ ## Digital Object Identifier (DOI) and access to dataset files
25
+
26
+ https://zenodo.org/record/4270158#.YTnXP0MzY0F
27
+
28
+ ## Introduction
29
+
30
+ TO DO: This is a dataset for Named Entity Recognition (NER) from...
31
+
32
+ ### Supported Tasks and Leaderboards
33
+
34
+ Named Entities Recognition, Language Model
35
+
36
+ ### Languages
37
+
38
+ ES - Spanish
39
+
40
+ ### Directory structure
41
+
42
+ * pharmaconer.py
43
+ * dev.conll
44
+ * test.conll
45
+ * train.conll
46
+ * README.md
47
+
48
+ ## Dataset Structure
49
+
50
+ ### Data Instances
51
+
52
+ Three four-column files, one for each split.
53
+
54
+ ### Data Fields
55
+
56
+ Every file has four columns:
57
+ * 1st column: Word form or punctuation symbol
58
+ * 2nd column: Original BRAT file name
59
+ * 3rd column: Spans
60
+ * 4th column: IOB tag
61
+
62
+ ### Example:
63
+ <pre>
64
+ La S0004-06142006000900008-1 123_125 O
65
+ paciente S0004-06142006000900008-1 126_134 O
66
+ tenía S0004-06142006000900008-1 135_140 O
67
+ antecedentes S0004-06142006000900008-1 141_153 O
68
+ de S0004-06142006000900008-1 154_156 O
69
+ hipotiroidismo S0004-06142006000900008-1 157_171 O
70
+ , S0004-06142006000900008-1 171_172 O
71
+ hipertensión S0004-06142006000900008-1 173_185 O
72
+ arterial S0004-06142006000900008-1 186_194 O
73
+ en S0004-06142006000900008-1 195_197 O
74
+ tratamiento S0004-06142006000900008-1 198_209 O
75
+ habitual S0004-06142006000900008-1 210_218 O
76
+ con S0004-06142006000900008-1 219-222 O
77
+ atenolol S0004-06142006000900008-1 223_231 B-NORMALIZABLES
78
+ y S0004-06142006000900008-1 232_233 O
79
+ enalapril S0004-06142006000900008-1 234_243 B-NORMALIZABLES
80
+ </pre>
81
+
82
+ ### Data Splits
83
+
84
+ * train: 8,074 tokens
85
+ * development: 3,764 tokens
86
+ * test: 3,931 tokens
87
+
88
+ ## Dataset Creation
89
+
90
+ ### Methodology
91
+
92
+ TO DO
93
+
94
+ ### Curation Rationale
95
+
96
+ For compatibility with similar datasets in other languages, we followed as close as possible existing curation guidelines.
97
+
98
+ ### Source Data
99
+
100
+ #### Initial Data Collection and Normalization
101
+
102
+ TO DO
103
+
104
+ #### Who are the source language producers?
105
+
106
+ TO DO
107
+
108
+ ### Annotations
109
+
110
+ #### Annotation process
111
+
112
+ TO DO
113
+
114
+ #### Who are the annotators?
115
+
116
+ TO DO
117
+
118
+ ### Dataset Curators
119
+
120
+ TO DO: Martin?
121
+
122
+ ### Personal and Sensitive Information
123
+
124
+ No personal or sensitive information included.
125
+
126
+ ## Contact
127
+
128
+ TO DO: Casimiro?
129
+
130
+ ## License
131
+
132
+ <a rel="license" href="https://creativecommons.org/licenses/by/4.0/"><img alt="Attribution 4.0 International License" style="border-width:0" src="https://chriszabriskie.com/img/cc-by.png" width="100"/></a><br />This work is licensed under a <a rel="license" href="https://creativecommons.org/licenses/by/4.0/">Attribution 4.0 International License</a>.
dev.conll ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:928f7286deba5b62c27d34c385fea08ff37c042870ad449ee861dbc27962e2f4
3
+ size 4216843
pharmaconer.py ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Loading script for the PharmaCoNER dataset.
2
+ import datasets
3
+
4
+
5
+ logger = datasets.logging.get_logger(__name__)
6
+
7
+
8
+ _CITATION = """\
9
+ @inproceedings{gonzalez-agirre-etal-2019-pharmaconer,
10
+ title = "{P}harma{C}o{NER}: Pharmacological Substances, Compounds and proteins Named Entity Recognition track",
11
+ author = "Gonzalez-Agirre, Aitor and
12
+ Marimon, Montserrat and
13
+ Intxaurrondo, Ander and
14
+ Rabal, Obdulia and
15
+ Villegas, Marta and
16
+ Krallinger, Martin",
17
+ booktitle = "Proceedings of The 5th Workshop on BioNLP Open Shared Tasks",
18
+ month = nov,
19
+ year = "2019",
20
+ address = "Hong Kong, China",
21
+ publisher = "Association for Computational Linguistics",
22
+ url = "https://aclanthology.org/D19-5701",
23
+ doi = "10.18653/v1/D19-5701",
24
+ pages = "1--10",
25
+ abstract = "One of the biomedical entity types of relevance for medicine or biosciences are chemical compounds and drugs. The correct detection these entities is critical for other text mining applications building on them, such as adverse drug-reaction detection, medication-related fake news or drug-target extraction. Although a significant effort was made to detect mentions of drugs/chemicals in English texts, so far only very limited attempts were made to recognize them in medical documents in other languages. Taking into account the growing amount of medical publications and clinical records written in Spanish, we have organized the first shared task on detecting drug and chemical entities in Spanish medical documents. Additionally, we included a clinical concept-indexing sub-track asking teams to return SNOMED-CT identifiers related to drugs/chemicals for a collection of documents. For this task, named PharmaCoNER, we generated annotation guidelines together with a corpus of 1,000 manually annotated clinical case studies. A total of 22 teams participated in the sub-track 1, (77 system runs), and 7 teams in the sub-track 2 (19 system runs). Top scoring teams used sophisticated deep learning approaches yielding very competitive results with F-measures above 0.91. These results indicate that there is a real interest in promoting biomedical text mining efforts beyond English. We foresee that the PharmaCoNER annotation guidelines, corpus and participant systems will foster the development of new resources for clinical and biomedical text mining systems of Spanish medical data.",
26
+ }
27
+ """
28
+
29
+ _DESCRIPTION = """\
30
+ https://temu.bsc.es/pharmaconer/
31
+ """
32
+
33
+ _URL = "https://huggingface.co/datasets/BSC-TeMU/pharmaconer/resolve/main/"
34
+ # _URL = "./"
35
+ _TRAINING_FILE = "train.conll"
36
+ _DEV_FILE = "dev.conll"
37
+ _TEST_FILE = "test.conll"
38
+
39
+ class PharmaCoNERConfig(datasets.BuilderConfig):
40
+ """BuilderConfig for PharmaCoNER dataset"""
41
+
42
+ def __init__(self, **kwargs):
43
+ """BuilderConfig for PharmaCoNER.
44
+
45
+ Args:
46
+ **kwargs: keyword arguments forwarded to super.
47
+ """
48
+ super(PharmaCoNERConfig, self).__init__(**kwargs)
49
+
50
+
51
+ class PharmaCoNER(datasets.GeneratorBasedBuilder):
52
+ """PharmaCoNER dataset."""
53
+
54
+ BUILDER_CONFIGS = [
55
+ PharmaCoNERConfig(
56
+ name="PharmaCoNER",
57
+ version=datasets.Version("1.0.0"),
58
+ description="PharmaCoNER dataset"),
59
+ ]
60
+
61
+ def _info(self):
62
+ return datasets.DatasetInfo(
63
+ description=_DESCRIPTION,
64
+ features=datasets.Features(
65
+ {
66
+ "id": datasets.Value("string"),
67
+ "tokens": datasets.Sequence(datasets.Value("string")),
68
+ "ner_tags": datasets.Sequence(
69
+ datasets.features.ClassLabel(
70
+ names=[
71
+ "O",
72
+ "B-NO_NORMALIZABLES",
73
+ "B-NORMALIZABLES",
74
+ "B-PROTEINAS",
75
+ "B-UNCLEAR",
76
+ "I-NO_NORMALIZABLES",
77
+ "I-NORMALIZABLES",
78
+ "I-PROTEINAS",
79
+ "I-UNCLEAR",
80
+ ]
81
+ )
82
+ ),
83
+ }
84
+ ),
85
+ supervised_keys=None,
86
+ homepage="https://temu.bsc.es/pharmaconer/",
87
+ citation=_CITATION,
88
+ )
89
+
90
+ def _split_generators(self, dl_manager):
91
+ """Returns SplitGenerators."""
92
+ urls_to_download = {
93
+ "train": f"{_URL}{_TRAINING_FILE}",
94
+ "dev": f"{_URL}{_DEV_FILE}",
95
+ "test": f"{_URL}{_TEST_FILE}",
96
+ }
97
+ downloaded_files = dl_manager.download_and_extract(urls_to_download)
98
+
99
+ return [
100
+ datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
101
+ datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
102
+ datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
103
+ ]
104
+
105
+ def _generate_examples(self, filepath):
106
+ logger.info("⏳ Generating examples from = %s", filepath)
107
+ with open(filepath, encoding="utf-8") as f:
108
+ guid = 0
109
+ tokens = []
110
+ pos_tags = []
111
+ ner_tags = []
112
+ for line in f:
113
+ if line.startswith("-DOCSTART-") or line == "" or line == "\n":
114
+ if tokens:
115
+ yield guid, {
116
+ "id": str(guid),
117
+ "tokens": tokens,
118
+ "ner_tags": ner_tags,
119
+ }
120
+ guid += 1
121
+ tokens = []
122
+ ner_tags = []
123
+ else:
124
+ # PharmaCoNER tokens are tab separated
125
+ splits = line.split("\t")
126
+ tokens.append(splits[0])
127
+ ner_tags.append(splits[-1].rstrip())
128
+ # last example
129
+ yield guid, {
130
+ "id": str(guid),
131
+ "tokens": tokens,
132
+ "ner_tags": ner_tags,
133
+ }
test.conll ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1be9198db6f1d6ff24be10fdb95102e866f0bd462b804057f9a28a708003b37
3
+ size 4386177
train.conll ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:675aa1b5a2a5424d44c0dcaa197227b61bc8ef11cd0099f9ded0564a9bc59007
3
+ size 8830998