chintagunta85
commited on
Commit
•
1c90f86
1
Parent(s):
346ae68
Upload pv_dataset.py
Browse files- pv_dataset.py +112 -0
pv_dataset.py
ADDED
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import datasets
|
2 |
+
import json
|
3 |
+
|
4 |
+
logger = datasets.logging.get_logger(__name__)
|
5 |
+
|
6 |
+
|
7 |
+
_HOMEPAGE = "https://www.google.com"
|
8 |
+
_TRAINING_FILE = "pv_train.tsv"
|
9 |
+
_DEV_FILE = "pv_val.tsv"
|
10 |
+
_TEST_FILE = "pv_test.tsv"
|
11 |
+
|
12 |
+
|
13 |
+
class PVDatasetConfig(datasets.BuilderConfig):
|
14 |
+
"""BuilderConfig for Bc2gmCorpus"""
|
15 |
+
|
16 |
+
def __init__(self, **kwargs):
|
17 |
+
"""BuilderConfig for Bc2gmCorpus.
|
18 |
+
Args:
|
19 |
+
**kwargs: keyword arguments forwarded to super.
|
20 |
+
"""
|
21 |
+
super(PVDatasetConfig, self).__init__(**kwargs)
|
22 |
+
|
23 |
+
class PVDataset(datasets.GeneratorBasedBuilder):
|
24 |
+
"""Bc2gmCorpus dataset."""
|
25 |
+
|
26 |
+
BUILDER_CONFIGS = [
|
27 |
+
PVDatasetConfig(name="PVDatasetCorpus", version=datasets.Version("1.0.0"), description="PVDataset"),
|
28 |
+
]
|
29 |
+
|
30 |
+
def _info(self):
|
31 |
+
|
32 |
+
custom_names = ['O','B-GENE','I-GENE','B-CHEMICAL','I-CHEMICAL','B-DISEASE','I-DISEASE','B-SPECIES', 'I-SPECIES']
|
33 |
+
|
34 |
+
return datasets.DatasetInfo(
|
35 |
+
description='abhi',
|
36 |
+
features=datasets.Features(
|
37 |
+
{
|
38 |
+
"id": datasets.Value("string"),
|
39 |
+
"tokens": datasets.Sequence(datasets.Value("string")),
|
40 |
+
"ner_tags": datasets.Sequence(
|
41 |
+
datasets.features.ClassLabel(
|
42 |
+
names=custom_names
|
43 |
+
)
|
44 |
+
),
|
45 |
+
}
|
46 |
+
),
|
47 |
+
supervised_keys=None,
|
48 |
+
homepage=_HOMEPAGE,
|
49 |
+
citation='cite me',
|
50 |
+
)
|
51 |
+
|
52 |
+
def _split_generators(self, dl_manager):
|
53 |
+
"""Returns SplitGenerators."""
|
54 |
+
urls_to_download = {
|
55 |
+
"train": f"{_TRAINING_FILE}",
|
56 |
+
"dev": f"{_DEV_FILE}",
|
57 |
+
"test": f"{_TEST_FILE}",
|
58 |
+
}
|
59 |
+
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
60 |
+
|
61 |
+
return [
|
62 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
|
63 |
+
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
|
64 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
|
65 |
+
]
|
66 |
+
|
67 |
+
def _generate_examples(self, filepath):
|
68 |
+
shift = 0
|
69 |
+
logger.info("⏳ Generating examples from = %s", filepath)
|
70 |
+
with open(filepath, encoding="utf-8") as f:
|
71 |
+
guid = 0
|
72 |
+
tokens = []
|
73 |
+
ner_tags = []
|
74 |
+
for line in f:
|
75 |
+
if line == "" or line == "\n":
|
76 |
+
if tokens:
|
77 |
+
yield guid, {
|
78 |
+
"id": str(guid),
|
79 |
+
"tokens": tokens,
|
80 |
+
"ner_tags": ner_tags,
|
81 |
+
}
|
82 |
+
guid += 1
|
83 |
+
tokens = []
|
84 |
+
ner_tags = []
|
85 |
+
else:
|
86 |
+
# tokens are tab separated
|
87 |
+
splits = line.split("\t")
|
88 |
+
tokens.append(splits[0])
|
89 |
+
if(splits[1].rstrip()=="B"):
|
90 |
+
ner_tags.append("B-SPECIES")
|
91 |
+
elif(splits[1].rstrip()=="I"):
|
92 |
+
ner_tags.append("I-SPECIES")
|
93 |
+
elif(splits[1].rstrip()=="B-Chemical"):
|
94 |
+
ner_tags.append("B-CHEMICAL")
|
95 |
+
elif(splits[1].rstrip()=="I-Chemical"):
|
96 |
+
ner_tags.append("I-CHEMICAL")
|
97 |
+
elif(splits[1].rstrip()=="B-Disease"):
|
98 |
+
ner_tags.append("B-DISEASE")
|
99 |
+
elif(splits[1].rstrip()=="I-Disease"):
|
100 |
+
ner_tags.append("I-DISEASE")
|
101 |
+
else:
|
102 |
+
ner_tags.append(splits[1].rstrip())
|
103 |
+
# last example
|
104 |
+
yield guid, {
|
105 |
+
"id": str(guid),
|
106 |
+
"tokens": tokens,
|
107 |
+
"ner_tags": ner_tags,
|
108 |
+
}
|
109 |
+
|
110 |
+
|
111 |
+
|
112 |
+
|