Datasets:

Languages:
Chavacano
ArXiv:
License:
holylovenia commited on
Commit
1a90928
1 Parent(s): edf66af

Upload creole_rc.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. creole_rc.py +207 -0
creole_rc.py ADDED
@@ -0,0 +1,207 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+
16
+ import csv
17
+ import json
18
+ from pathlib import Path
19
+ from typing import Dict, List, Tuple
20
+
21
+ import datasets
22
+
23
+ from seacrowd.utils import schemas
24
+ from seacrowd.utils.configs import SEACrowdConfig
25
+ from seacrowd.utils.constants import Licenses, Tasks
26
+
27
+ _CITATION = """\
28
+ @misc{lent2023creoleval,
29
+ title={CreoleVal: Multilingual Multitask Benchmarks for Creoles},
30
+ author={Heather Lent and Kushal Tatariya and Raj Dabre and Yiyi Chen and Marcell Fekete and Esther Ploeger and Li Zhou and Hans Erik Heje and Diptesh Kanojia and Paul Belony and Marcel Bollmann and \
31
+ Loïc Grobol and Miryam de Lhoneux and Daniel Hershcovich and Michel DeGraff and Anders Søgaard and Johannes Bjerva},
32
+ year={2023},
33
+ eprint={2310.19567},
34
+ archivePrefix={arXiv},
35
+ primaryClass={cs.CL}
36
+ }
37
+ """
38
+
39
+ _DATASETNAME = "creole_rc"
40
+ _DESCRIPTION = """\
41
+ CreoleRC is a subset created by the CreoleVal paper. Relation classification (RC) aims to identify semantic associations between entities within a text, essential for applications like knowledge base \
42
+ completion and question answering. The dataset is sourced from Wikipedia and manually annotated. CreoleRC contains 5 creoles, but SEACrowd is interested specifically in the Chavacano subset.
43
+ """
44
+ _HOMEPAGE = "https://github.com/hclent/CreoleVal/tree/main/nlu/relation_classification"
45
+ _LANGUAGES = ["cbk"]
46
+ _LICENSE = Licenses.CC_BY_SA_4_0.value
47
+ _LOCAL = False
48
+ _URLS = {
49
+ "csv": "https://raw.githubusercontent.com/hclent/CreoleVal/main/nlu/relation_classification/data/relation_extraction/cbk-zam.csv",
50
+ "json": "https://raw.githubusercontent.com/hclent/CreoleVal/main/nlu/relation_classification/data/relation_extraction/cbk-zam.json",
51
+ }
52
+ _SUPPORTED_TASKS = [Tasks.RELATION_EXTRACTION]
53
+
54
+ _SOURCE_VERSION = "1.0.0"
55
+ _SEACROWD_VERSION = "2024.06.20"
56
+
57
+
58
+ class CreoleRC(datasets.GeneratorBasedBuilder):
59
+ """Creole relation classification dataset, Chavacano subset, from https://github.com/hclent/CreoleVal/tree/main/nlu/relation_classification."""
60
+
61
+ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
62
+ SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
63
+
64
+ # You will be able to load the "source" or "seacrowd" configurations with
65
+ # ds_source = datasets.load_dataset('my_dataset', name='source')
66
+ # ds_seacrowd = datasets.load_dataset('my_dataset', name='seacrowd')
67
+
68
+ # For local datasets you can make use of the `data_dir` and `data_files` kwargs
69
+ # https://huggingface.co/docs/datasets/add_dataset.html#downloading-data-files-and-organizing-splits
70
+ # ds_source = datasets.load_dataset('my_dataset', name='source', data_dir="/path/to/data/files")
71
+ # ds_seacrowd = datasets.load_dataset('my_dataset', name='seacrowd', data_dir="/path/to/data/files")
72
+
73
+ SEACROWD_SCHEMA_NAME = "kb"
74
+
75
+ BUILDER_CONFIGS = [
76
+ SEACrowdConfig(
77
+ name=f"{_DATASETNAME}_source",
78
+ version=SOURCE_VERSION,
79
+ description=f"{_DATASETNAME} source schema",
80
+ schema="source",
81
+ subset_id=_DATASETNAME,
82
+ ),
83
+ SEACrowdConfig(
84
+ name=f"{_DATASETNAME}_seacrowd_{SEACROWD_SCHEMA_NAME}",
85
+ version=SEACROWD_VERSION,
86
+ description=f"{_DATASETNAME} SEACrowd schema",
87
+ schema=f"seacrowd_{SEACROWD_SCHEMA_NAME}",
88
+ subset_id=_DATASETNAME,
89
+ ),
90
+ ]
91
+
92
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
93
+
94
+ def _info(self) -> datasets.DatasetInfo:
95
+
96
+ if self.config.schema == "source":
97
+ features = datasets.Features(
98
+ {
99
+ "id": datasets.Value("string"),
100
+ "sentence": datasets.Value("string"),
101
+ "ent1": datasets.Value("string"),
102
+ "ent2": datasets.Value("string"),
103
+ "ent1_qcode": datasets.Value("string"),
104
+ "ent2_qcode": datasets.Value("string"),
105
+ "property": datasets.Value("string"),
106
+ "property_desc": datasets.Value("string"),
107
+ "tokens": datasets.Sequence(datasets.Value("string")),
108
+ "edgeset_left": datasets.Sequence(datasets.Value("int32")),
109
+ "edgeset_right": datasets.Sequence(datasets.Value("int32")),
110
+ "edgeset_triple": datasets.Sequence(datasets.Value("string")),
111
+ }
112
+ )
113
+
114
+ elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
115
+ features = schemas.kb_features
116
+
117
+ return datasets.DatasetInfo(
118
+ description=_DESCRIPTION,
119
+ features=features,
120
+ homepage=_HOMEPAGE,
121
+ license=_LICENSE,
122
+ citation=_CITATION,
123
+ )
124
+
125
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
126
+ """Returns SplitGenerators."""
127
+
128
+ data_paths = {
129
+ "csv": Path(dl_manager.download_and_extract(_URLS["csv"])),
130
+ "json": Path(dl_manager.download_and_extract(_URLS["json"])),
131
+ }
132
+
133
+ return [
134
+ datasets.SplitGenerator(
135
+ name=datasets.Split.TRAIN,
136
+ gen_kwargs={
137
+ "csv_filepath": data_paths["csv"],
138
+ "json_filepath": data_paths["json"],
139
+ "split": "train",
140
+ },
141
+ )
142
+ ]
143
+
144
+ # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
145
+
146
+ def _generate_examples(self, csv_filepath: Path, json_filepath: Path, split: str) -> Tuple[int, Dict]:
147
+ """Yields examples as (key, example) tuples."""
148
+ # read csv file
149
+ with open(csv_filepath, "r", encoding="utf-8") as csv_file:
150
+ csv_reader = csv.reader(csv_file)
151
+ csv_data = [row for row in csv_reader]
152
+ csv_data = csv_data[1:] # remove header
153
+
154
+ # read json file
155
+ with open(json_filepath, "r", encoding="utf-8") as json_file:
156
+ json_data = json.load(json_file)
157
+
158
+ # properties descriptions from https://github.com/hclent/CreoleVal/tree/main/nlu/relation_classification
159
+ # for properties present in Chavacano subset
160
+ properties_desc = {"P17": "country", "P30": "continent", "P106": "occupation", "P131": "located in the administrative territorial entity", "P361": "part of", "P1376": " capital of country"}
161
+
162
+ num_sample = len(csv_data)
163
+
164
+ for i in range(num_sample):
165
+ if self.config.schema == "source":
166
+ example = {
167
+ "id": str(i),
168
+ "sentence": csv_data[i][0],
169
+ "ent1": csv_data[i][1],
170
+ "ent2": csv_data[i][2],
171
+ "ent1_qcode": csv_data[i][3],
172
+ "ent2_qcode": csv_data[i][4],
173
+ "property": csv_data[i][5],
174
+ "property_desc": properties_desc[csv_data[i][5]],
175
+ "tokens": json_data[i]["tokens"],
176
+ "edgeset_left": json_data[i]["edgeSet"]["left"],
177
+ "edgeset_right": json_data[i]["edgeSet"]["right"],
178
+ "edgeset_triple": json_data[i]["edgeSet"]["triple"],
179
+ }
180
+ elif self.config.schema == f"seacrowd_{self.SEACROWD_SCHEMA_NAME}":
181
+ offset_entity1 = csv_data[i][0].find(csv_data[i][1])
182
+ offset_entity2 = csv_data[i][0].find(csv_data[i][2])
183
+
184
+ if (offset_entity1 == -1) or (offset_entity2 == -1):
185
+ continue
186
+
187
+ example = {
188
+ "id": str(i),
189
+ "passages": [{"id": f"passage-{i}", "type": "text", "text": [csv_data[i][0]], "offsets": [[0, len(csv_data[i][0])]]}],
190
+ "entities": [
191
+ {"id": f"{i}-entity-{csv_data[i][3]}", "type": "text", "text": [csv_data[i][1]], "normalized": [{"db_name": csv_data[i][1], "db_id": csv_data[i][3]}], "offsets": [[offset_entity1, offset_entity1 + len(csv_data[i][1])]]},
192
+ {"id": f"{i}-entity-{csv_data[i][4]}", "type": "text", "text": [csv_data[i][2]], "normalized": [{"db_name": csv_data[i][2], "db_id": csv_data[i][4]}], "offsets": [[offset_entity2, offset_entity2 + len(csv_data[i][2])]]},
193
+ ],
194
+ "events": [],
195
+ "coreferences": [],
196
+ "relations": [
197
+ {
198
+ "id": f"{i}-relation-{csv_data[i][5]}",
199
+ "type": properties_desc[csv_data[i][5]],
200
+ "arg1_id": f"{i}-entity-{csv_data[i][3]}",
201
+ "arg2_id": f"{i}-entity-{csv_data[i][4]}",
202
+ "normalized": [{"db_name": properties_desc[csv_data[i][5]], "db_id": csv_data[i][5]}],
203
+ }
204
+ ],
205
+ }
206
+
207
+ yield i, example