Datasets:
ArXiv:
Fehler in lener-br behoben
Browse files- legalglue.py +76 -6
legalglue.py
CHANGED
@@ -84,6 +84,32 @@ LENER_BR=[
|
|
84 |
"I-JURISPRUDENCIA",
|
85 |
]
|
86 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
class LegalGlueConfig(datasets.BuilderConfig):
|
88 |
"""BuilderConfig for LegalGLUE."""
|
89 |
|
@@ -95,6 +121,9 @@ class LegalGlueConfig(datasets.BuilderConfig):
|
|
95 |
citation, #citation for the dataset
|
96 |
data_url,
|
97 |
data_files,
|
|
|
|
|
|
|
98 |
**kwargs,
|
99 |
):
|
100 |
super(LegalGlueConfig, self).__init__(version=datasets.Version("1.1.0", ""), **kwargs)
|
@@ -104,6 +133,13 @@ class LegalGlueConfig(datasets.BuilderConfig):
|
|
104 |
self.citation = citation
|
105 |
self.data_url = data_url
|
106 |
self.data_files = data_files
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
107 |
|
108 |
|
109 |
|
@@ -142,7 +178,9 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
|
|
142 |
note = {10/11 September 2019},
|
143 |
pages = {272--287},
|
144 |
pdf = {https://link.springer.com/content/pdf/10.1007%2F978-3-030-33220-4_20.pdf}}
|
145 |
-
""")
|
|
|
|
|
146 |
),
|
147 |
LegalGlueConfig(
|
148 |
name="lener_br",
|
@@ -180,7 +218,8 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
|
|
180 |
doi = {10.1007/978-3-319-99722-3_32},
|
181 |
url = {https://cic.unb.br/~teodecampos/LeNER-Br/},
|
182 |
}
|
183 |
-
""")
|
|
|
184 |
),
|
185 |
LegalGlueConfig(
|
186 |
name="cuad",
|
@@ -203,8 +242,39 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
|
|
203 |
journal={arXiv preprint arXiv:2103.06268},
|
204 |
year={2021}
|
205 |
}
|
206 |
-
""")
|
207 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
208 |
]
|
209 |
|
210 |
def _info(self):
|
@@ -401,7 +471,7 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
|
|
401 |
}
|
402 |
id += 1
|
403 |
tokens = []
|
404 |
-
|
405 |
else:
|
406 |
token, tag = line.split()
|
407 |
tokens.append(token)
|
@@ -410,7 +480,7 @@ class LegalGLUE(datasets.GeneratorBasedBuilder):
|
|
410 |
yield id, {
|
411 |
"id": str(id),
|
412 |
"tokens": tokens,
|
413 |
-
"ner_tags":
|
414 |
}
|
415 |
|
416 |
elif self.config.name == "cuad":
|
|
|
84 |
"I-JURISPRUDENCIA",
|
85 |
]
|
86 |
|
87 |
+
# MULTI_EURLEX_LANGUAGES = [
|
88 |
+
# "en",
|
89 |
+
# "da",
|
90 |
+
# "de",
|
91 |
+
# "nl",
|
92 |
+
# "sv",
|
93 |
+
# "bg",
|
94 |
+
# "cs",
|
95 |
+
# "hr",
|
96 |
+
# "pl",
|
97 |
+
# "sk",
|
98 |
+
# "sl",
|
99 |
+
# "es",
|
100 |
+
# "fr",
|
101 |
+
# "it",
|
102 |
+
# "pt",
|
103 |
+
# "ro",
|
104 |
+
# "et",
|
105 |
+
# "fi",
|
106 |
+
# "hu",
|
107 |
+
# "lt",
|
108 |
+
# "lv",
|
109 |
+
# "el",
|
110 |
+
# "mt",
|
111 |
+
# ]
|
112 |
+
|
113 |
class LegalGlueConfig(datasets.BuilderConfig):
|
114 |
"""BuilderConfig for LegalGLUE."""
|
115 |
|
|
|
121 |
citation, #citation for the dataset
|
122 |
data_url,
|
123 |
data_files,
|
124 |
+
# language,
|
125 |
+
# languages = None,
|
126 |
+
# label_level = None,
|
127 |
**kwargs,
|
128 |
):
|
129 |
super(LegalGlueConfig, self).__init__(version=datasets.Version("1.1.0", ""), **kwargs)
|
|
|
133 |
self.citation = citation
|
134 |
self.data_url = data_url
|
135 |
self.data_files = data_files
|
136 |
+
# self.language = language
|
137 |
+
# self.label_level = label_level
|
138 |
+
#
|
139 |
+
# if language != "all_languages":
|
140 |
+
# self.languages = [language]
|
141 |
+
# else:
|
142 |
+
# self.languages = languages if languages is not None else MULTI_EURLEX_LANGUAGES
|
143 |
|
144 |
|
145 |
|
|
|
178 |
note = {10/11 September 2019},
|
179 |
pages = {272--287},
|
180 |
pdf = {https://link.springer.com/content/pdf/10.1007%2F978-3-030-33220-4_20.pdf}}
|
181 |
+
"""),
|
182 |
+
# language= "German"
|
183 |
+
|
184 |
),
|
185 |
LegalGlueConfig(
|
186 |
name="lener_br",
|
|
|
218 |
doi = {10.1007/978-3-319-99722-3_32},
|
219 |
url = {https://cic.unb.br/~teodecampos/LeNER-Br/},
|
220 |
}
|
221 |
+
"""),
|
222 |
+
# language = "Portuguese"
|
223 |
),
|
224 |
LegalGlueConfig(
|
225 |
name="cuad",
|
|
|
242 |
journal={arXiv preprint arXiv:2103.06268},
|
243 |
year={2021}
|
244 |
}
|
245 |
+
"""),
|
246 |
+
# language = "English"
|
247 |
+
),
|
248 |
+
# LegalGlueConfig(
|
249 |
+
# name="multi_eurlex",
|
250 |
+
# description=textwrap.dedent(
|
251 |
+
# """\
|
252 |
+
# MultiEURLEX comprises 65k EU laws in 23 official EU languages (some low-ish resource).
|
253 |
+
# Each EU law has been annotated with EUROVOC concepts (labels) by the Publication Office of EU.
|
254 |
+
# As with the English EURLEX, the goal is to predict the relevant EUROVOC concepts (labels);
|
255 |
+
# this is multi-label classification task (given the text, predict multiple labels).
|
256 |
+
# """
|
257 |
+
# ),
|
258 |
+
# label_classes=None,
|
259 |
+
# multi_label=True,
|
260 |
+
# data_url="https://zenodo.org/record/5363165/files/multi_eurlex.tar.gz",
|
261 |
+
# data_files=["train.json", "test.json", "dev.json"],
|
262 |
+
# # homepage="https://www.atticusprojectai.org/cuad",
|
263 |
+
# citation=textwrap.dedent("""\
|
264 |
+
# @InProceedings{chalkidis-etal-2021-multieurlex,
|
265 |
+
# author = {Chalkidis, Ilias
|
266 |
+
# and Fergadiotis, Manos
|
267 |
+
# and Androutsopoulos, Ion},
|
268 |
+
# title = {MultiEURLEX -- A multi-lingual and multi-label legal document
|
269 |
+
# classification dataset for zero-shot cross-lingual transfer},
|
270 |
+
# booktitle = {Proceedings of the 2021 Conference on Empirical Methods
|
271 |
+
# in Natural Language Processing},
|
272 |
+
# year = {2021},
|
273 |
+
# publisher = {Association for Computational Linguistics},
|
274 |
+
# location = {Punta Cana, Dominican Republic},
|
275 |
+
# }"""
|
276 |
+
# )
|
277 |
+
# )
|
278 |
]
|
279 |
|
280 |
def _info(self):
|
|
|
471 |
}
|
472 |
id += 1
|
473 |
tokens = []
|
474 |
+
tags = []
|
475 |
else:
|
476 |
token, tag = line.split()
|
477 |
tokens.append(token)
|
|
|
480 |
yield id, {
|
481 |
"id": str(id),
|
482 |
"tokens": tokens,
|
483 |
+
"ner_tags": tags,
|
484 |
}
|
485 |
|
486 |
elif self.config.name == "cuad":
|