Datasets:

Modalities:
Text
ArXiv:
Libraries:
Datasets
License:
File size: 13,126 Bytes
c3db668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
568ffc1
 
 
 
c3db668
 
 
 
 
 
 
 
 
568ffc1
 
 
 
c3db668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
568ffc1
 
 
c3db668
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
568ffc1
 
c3db668
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
# coding=utf-8
# Copyright 2020 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3
"""REDFM: a Filtered and Multilingual Relation Extraction Dataset."""


import collections
import json
import os
from contextlib import ExitStack
import logging
import datasets


_CITATION = """\
@InProceedings{REDFM2023,
  author = {Huguet Cabot, Pere-Lluis
                 and Tedeschi, Simone
                 and Ngonga Ngomo, Axel-Cyrille
                 and Navigli, Roberto},
  title = {RED\textsuperscript{FM}: a Filtered and Multilingual Relation Extraction Dataset},
  booktitle = {Proceedings of the 2023 Conference on Association for Computational Linguistics},
  year = {2023},
  publisher = {Association for Computational Linguistics},
  location = {Toronto, Canada},
}"""

_DESCRIPTION = """\
Relation Extraction (RE) is a task that identifies relationships between entities in a text, enabling the acquisition of relational facts and bridging the gap between natural language and structured knowledge. However, current RE models often rely on small datasets with low coverage of relation types, particularly when working with languages other than English. \\
In this paper, we address the above issue and provide two new resources that enable the training and evaluation of multilingual RE systems.
First, we present SRED\textsuperscript{FM}, an automatically annotated dataset covering 18 languages, 400 relation types, 13 entity types, totaling more than 40 million triplet instances. Second, we propose RED\textsuperscript{FM}, a smaller, human-revised dataset for seven languages that allows for the evaluation of multilingual RE systems. 
To demonstrate the utility of these novel datasets, we experiment with the first end-to-end multilingual RE model, mREBEL, 
that extracts triplets, including entity types, in multiple languages. We release our resources and model checkpoints at \href{https://www.github.com/babelscape/rebel}{https://www.github.com/babelscape/rebel}. 
"""

DEFAULT_CONFIG_NAME = "all_languages"

_LANGUAGES = ("ar", "ca", "de", "el", "en", "es", "fr", "hi", "it", "ja", "ko", "nl", "pl", "pt", "ru", "sv", "vi", "zh")

_URL_train = f"data/train."
_URL_dev = f"data/dev." 
_URL_test = f"data/test."

class SREDFMConfig(datasets.BuilderConfig):
    """BuilderConfig for SREDFM."""

    def __init__(self, language: str, languages=None, **kwargs):
        """BuilderConfig for SREDFM.
        Args:
        language: One of ar,de,en,es,fr,it,zh, or all_languages
          **kwargs: keyword arguments forwarded to super.
        """
        super(SREDFMConfig, self).__init__(**kwargs)
        self.language = language
        if language != "all_languages":
            self.languages = [language]
        else:
            self.languages = languages if languages is not None else _LANGUAGES


class SREDFM(datasets.GeneratorBasedBuilder):
    """SREDFM: a Filtered and Multilingual Relation Extraction Dataset. Version 1.0.0"""

    VERSION = datasets.Version("1.0.0", "")
    BUILDER_CONFIG_CLASS = SREDFMConfig
    BUILDER_CONFIGS = [
        SREDFMConfig(
            name=lang,
            language=lang,
            version=datasets.Version("1.0.0", ""),
            description=f"Plain text import of SREDFM for the {lang} language",
        )
        for lang in _LANGUAGES
    ] + [
        SREDFMConfig(
            name="all_languages",
            language="all_languages",
            version=datasets.Version("1.0.0", ""),
            description="Plain text import of SREDFM for all languages",
        )
    ]

    def _info(self):
        if self.config.language == "all_languages":
            features = datasets.Features(
                {
                    "docid": datasets.Value("string"),
                    "title": datasets.Value("string"),
                    "uri": datasets.Value("string"),
                    "lan": datasets.Value("string"),
                    "text": datasets.Value("string"),
                    "entities": [{'uri': datasets.Value(dtype='string'), 'surfaceform': datasets.Value(dtype='string'), 'type': datasets.Value(dtype='string'), 'start': datasets.Value(dtype='int32'), 'end': datasets.Value(dtype='int32')}],
                    "relations": [{'subject': datasets.Value(dtype='int32'),
                                    'predicate': datasets.Value(dtype='string'),
                                    'object': datasets.Value(dtype='int32')}],
                }
            )
        else:
            features = datasets.Features(
                {
                    "docid": datasets.Value("string"),
                    "title": datasets.Value("string"),
                    "uri": datasets.Value("string"),
                    "text": datasets.Value("string"),
                    "entities": [{'uri': datasets.Value(dtype='string'), 'surfaceform': datasets.Value(dtype='string'), 'type': datasets.Value(dtype='string'), 'start': datasets.Value(dtype='int32'), 'end': datasets.Value(dtype='int32')}],
                    "relations": [{'subject': datasets.Value(dtype='int32'),
                                    'predicate': datasets.Value(dtype='string'),
                                    'object': datasets.Value(dtype='int32')}],
                }
            )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            # No default supervised_keys (as we have to pass both premise
            # and hypothesis as input).
            supervised_keys=None,
            homepage="https://www.github.com/babelscape/rebel",
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        data_dir = dl_manager.download(
            {
                "train": [f"{_URL_train}{lang}.jsonl" for lang in self.config.languages],
                "dev": [f"{_URL_dev}{lang}.jsonl" for lang in self.config.languages],
                "test": [f"{_URL_test}{lang}.jsonl" for lang in self.config.languages],
                "relations": "relations.tsv",
            }
        )

        return [            
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "filepaths": data_dir["train"],
                    "relations": data_dir["relations"],
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={
                    "filepaths": data_dir["test"],
                    "relations": data_dir["relations"],
                },
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={
                    "filepaths": data_dir["dev"],
                    "relations": data_dir["relations"],
                },
            ),
        ]

    def _generate_examples(self, relations, filepaths):
        """This function returns the examples in the raw (text) form."""
        logging.info("generating examples from = %s", filepaths)
        relation_names = dict()
        with open(relations, encoding="utf-8") as f:
            for row in f:
                rel_code, rel_name, rel_alt_names, rel_description = row.strip().split("\t")
                relation_names[rel_code] = rel_name
        if self.config.language == "all_languages":
            for filepath in filepaths:
                with open(filepath, encoding="utf-8") as f:
                    for idx, row in enumerate(f):
                        data = json.loads(row)
                        entities = []
                        for entity in data["entities"]:
                            entities.append({
                                "uri": entity["uri"],
                                "surfaceform": entity["surfaceform"],
                                "start": entity["boundaries"][0],
                                "end": entity["boundaries"][1],
                                "type": entity["type"],
                            })
                        relations = []
                        for relation in data["relations"]:
                            if relation["predicate"]["uri"] not in relation_names or relation['confidence']<=0.75:
                                continue
                            relations.append({
                                "subject": entities.index({
                                    "uri": relation["subject"]["uri"],
                                    "surfaceform": relation["subject"]["surfaceform"],
                                    "start": relation["subject"]["boundaries"][0],
                                    "end": relation["subject"]["boundaries"][1],
                                    "type": relation["subject"]["type"],
                                }),
                                "predicate": relation_names[relation["predicate"]["uri"]],
                                "object": entities.index({
                                    "uri": relation["object"]["uri"],
                                    "surfaceform": relation["object"]["surfaceform"],
                                    "start": relation["object"]["boundaries"][0],
                                    "end": relation["object"]["boundaries"][1],
                                    "type": relation["object"]["type"],
                                }),
                            })         
                        if len(relations) == 0:
                            continue               
                        yield data["docid"]+ '-' + data["lan"], {
                            "docid": data["docid"],
                            "title": data["title"],
                            "uri": data["uri"],
                            "lan": data["lan"],
                            "text": data["text"],
                            "entities": entities,
                            "relations": relations,
                        }
        else:
            for filepath in filepaths:
                with open(filepath, encoding="utf-8") as f:
                    for idx, row in enumerate(f):
                        data = json.loads(row)
                        entities = []
                        for entity in data["entities"]:
                            entities.append({
                                "uri": entity["uri"],
                                "surfaceform": entity["surfaceform"],
                                "start": entity["boundaries"][0],
                                "end": entity["boundaries"][1],
                                "type": entity["type"],
                            })
                        relations = []
                        for relation in data["relations"]:
                            if relation["predicate"]["uri"] not in relation_names or relation['confidence']<=0.75:
                                continue
                            relations.append({
                                "subject": entities.index({
                                    "uri": relation["subject"]["uri"],
                                    "surfaceform": relation["subject"]["surfaceform"],
                                    "start": relation["subject"]["boundaries"][0],
                                    "end": relation["subject"]["boundaries"][1],
                                    "type": relation["subject"]["type"],
                                }),
                                "predicate": relation_names[relation["predicate"]["uri"]],
                                "object": entities.index({
                                    "uri": relation["object"]["uri"],
                                    "surfaceform": relation["object"]["surfaceform"],
                                    "start": relation["object"]["boundaries"][0],
                                    "end": relation["object"]["boundaries"][1],
                                    "type": relation["object"]["type"],
                                }),
                            })
                        if len(relations) == 0:
                            continue     
                        yield data["docid"], {
                            "docid": data["docid"],
                            "title": data["title"],
                            "uri": data["uri"],
                            "text": data["text"],
                            "entities": entities,
                            "relations": relations,
                        }