Upload tlunified_ner.py with huggingface_hub
Browse files- tlunified_ner.py +149 -0
tlunified_ner.py
ADDED
@@ -0,0 +1,149 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from pathlib import Path
|
2 |
+
from typing import Dict, List, Tuple
|
3 |
+
|
4 |
+
import datasets
|
5 |
+
from datasets.download.download_manager import DownloadManager
|
6 |
+
|
7 |
+
from seacrowd.utils import schemas
|
8 |
+
from seacrowd.utils.configs import SEACrowdConfig
|
9 |
+
from seacrowd.utils.constants import Licenses, Tasks
|
10 |
+
|
11 |
+
_CITATION = """
|
12 |
+
@inproceedings{miranda-2023-developing,
|
13 |
+
title = {Developing a Named Entity Recognition Dataset for Tagalog},
|
14 |
+
author = "Miranda, Lester James Validad",
|
15 |
+
booktitle = "Proceedings of the First Workshop for Southeast Asian Language Processing (SEALP),"
|
16 |
+
month = nov,
|
17 |
+
year = 2023,
|
18 |
+
address = "Online",
|
19 |
+
publisher = "Association for Computational Linguistics",
|
20 |
+
}
|
21 |
+
"""
|
22 |
+
|
23 |
+
_LOCAL = False
|
24 |
+
_LANGUAGES = ["tgl"]
|
25 |
+
_DATASETNAME = "tlunified_ner"
|
26 |
+
_DESCRIPTION = """\
|
27 |
+
This dataset contains the annotated TLUnified corpora from Cruz and Cheng
|
28 |
+
(2021). It is a curated sample of around 7,000 documents for the named entity
|
29 |
+
recognition (NER) task. The majority of the corpus are news reports in Tagalog,
|
30 |
+
resembling the domain of the original ConLL 2003. There are three entity types:
|
31 |
+
Person (PER), Organization (ORG), and Location (LOC).
|
32 |
+
"""
|
33 |
+
|
34 |
+
_HOMEPAGE = "https://huggingface.co/ljvmiranda921/tlunified-ner"
|
35 |
+
_LICENSE = Licenses.GPL_3_0.value
|
36 |
+
_URLS = {
|
37 |
+
"train": "https://huggingface.co/datasets/ljvmiranda921/tlunified-ner/resolve/main/corpus/iob/train.iob",
|
38 |
+
"dev": "https://huggingface.co/datasets/ljvmiranda921/tlunified-ner/resolve/main/corpus/iob/dev.iob",
|
39 |
+
"test": "https://huggingface.co/datasets/ljvmiranda921/tlunified-ner/resolve/main/corpus/iob/test.iob",
|
40 |
+
}
|
41 |
+
|
42 |
+
_SUPPORTED_TASKS = [Tasks.NAMED_ENTITY_RECOGNITION]
|
43 |
+
_SOURCE_VERSION = "1.0.0"
|
44 |
+
_SEACROWD_VERSION = "2024.06.20"
|
45 |
+
|
46 |
+
|
47 |
+
class TLUnifiedNERDataset(datasets.GeneratorBasedBuilder):
|
48 |
+
"""Tagalog Named Entity Recognition dataset from https://huggingface.co/ljvmiranda921/tlunified-ner"""
|
49 |
+
|
50 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
51 |
+
SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
|
52 |
+
|
53 |
+
SEACROWD_SCHEMA_NAME = "seq_label"
|
54 |
+
LABEL_CLASSES = ["O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC"]
|
55 |
+
|
56 |
+
BUILDER_CONFIGS = [
|
57 |
+
SEACrowdConfig(
|
58 |
+
name=f"{_DATASETNAME}_source",
|
59 |
+
version=SOURCE_VERSION,
|
60 |
+
description=f"{_DATASETNAME} source schema",
|
61 |
+
schema="source",
|
62 |
+
subset_id=_DATASETNAME,
|
63 |
+
),
|
64 |
+
SEACrowdConfig(
|
65 |
+
name=f"{_DATASETNAME}_seacrowd_{SEACROWD_SCHEMA_NAME}",
|
66 |
+
version=SEACROWD_VERSION,
|
67 |
+
description=f"{_DATASETNAME} SEACrowd schema",
|
68 |
+
schema=f"seacrowd_{SEACROWD_SCHEMA_NAME}",
|
69 |
+
subset_id=_DATASETNAME,
|
70 |
+
),
|
71 |
+
]
|
72 |
+
|
73 |
+
DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
|
74 |
+
|
75 |
+
def _info(self) -> datasets.DatasetInfo:
|
76 |
+
if self.config.schema == "source":
|
77 |
+
features = datasets.Features(
|
78 |
+
{
|
79 |
+
"id": datasets.Value("string"),
|
80 |
+
"tokens": datasets.Sequence(datasets.Value("string")),
|
81 |
+
"ner_tags": datasets.Sequence(datasets.features.ClassLabel(names=self.LABEL_CLASSES)),
|
82 |
+
}
|
83 |
+
)
|
84 |
+
elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
|
85 |
+
features = schemas.seq_label_features(self.LABEL_CLASSES)
|
86 |
+
|
87 |
+
return datasets.DatasetInfo(
|
88 |
+
description=_DESCRIPTION,
|
89 |
+
features=features,
|
90 |
+
homepage=_HOMEPAGE,
|
91 |
+
license=_LICENSE,
|
92 |
+
citation=_CITATION,
|
93 |
+
)
|
94 |
+
|
95 |
+
def _split_generators(self, dl_manager: DownloadManager) -> List[datasets.SplitGenerator]:
|
96 |
+
"""Returns SplitGenerators."""
|
97 |
+
data_files = {
|
98 |
+
"train": Path(dl_manager.download_and_extract(_URLS["train"])),
|
99 |
+
"dev": Path(dl_manager.download_and_extract(_URLS["dev"])),
|
100 |
+
"test": Path(dl_manager.download_and_extract(_URLS["test"])),
|
101 |
+
}
|
102 |
+
|
103 |
+
return [
|
104 |
+
datasets.SplitGenerator(
|
105 |
+
name=datasets.Split.TRAIN,
|
106 |
+
gen_kwargs={"filepath": data_files["train"], "split": "train"},
|
107 |
+
),
|
108 |
+
datasets.SplitGenerator(
|
109 |
+
name=datasets.Split.VALIDATION,
|
110 |
+
gen_kwargs={"filepath": data_files["dev"], "split": "dev"},
|
111 |
+
),
|
112 |
+
datasets.SplitGenerator(
|
113 |
+
name=datasets.Split.TEST,
|
114 |
+
gen_kwargs={"filepath": data_files["test"], "split": "test"},
|
115 |
+
),
|
116 |
+
]
|
117 |
+
|
118 |
+
def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
|
119 |
+
"""Yield examples as (key, example) tuples"""
|
120 |
+
# The only difference between the source schema and the seacrowd seq_label schema is the dictionary keys.
|
121 |
+
# The implementation is the same.
|
122 |
+
label_key = "ner_tags" if self.config.schema == "source" else "labels"
|
123 |
+
with open(filepath, encoding="utf-8") as f:
|
124 |
+
guid = 0
|
125 |
+
tokens = []
|
126 |
+
ner_tags = []
|
127 |
+
for line in f:
|
128 |
+
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
129 |
+
if tokens:
|
130 |
+
yield guid, {
|
131 |
+
"id": str(guid),
|
132 |
+
"tokens": tokens,
|
133 |
+
label_key: ner_tags,
|
134 |
+
}
|
135 |
+
guid += 1
|
136 |
+
tokens = []
|
137 |
+
ner_tags = []
|
138 |
+
else:
|
139 |
+
# TLUnified-NER iob are separated by \t
|
140 |
+
token, ner_tag = line.split("\t")
|
141 |
+
tokens.append(token)
|
142 |
+
ner_tags.append(ner_tag.rstrip())
|
143 |
+
# Last example
|
144 |
+
if tokens:
|
145 |
+
yield guid, {
|
146 |
+
"id": str(guid),
|
147 |
+
"tokens": tokens,
|
148 |
+
label_key: ner_tags,
|
149 |
+
}
|