joelniklaus
commited on
Commit
·
f782401
1
Parent(s):
1b1bbfd
updated data loader to include update from pile of law
Browse files- .gitattributes +1 -0
- Multi_Legal_Pile.py +207 -66
- README.md +46 -51
.gitattributes
CHANGED
@@ -77,3 +77,4 @@ data/fr/caselaw/switzerland_entscheidsuche.jsonl.xz filter=lfs diff=lfs merge=lf
|
|
77 |
data/it/caselaw/switzerland_entscheidsuche.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
78 |
data/en/legislation/uk_uk_lex.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
79 |
data/hu/legislation/hungary_marcell.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
|
|
|
77 |
data/it/caselaw/switzerland_entscheidsuche.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
78 |
data/en/legislation/uk_uk_lex.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
79 |
data/hu/legislation/hungary_marcell.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
80 |
+
data/pt/caselaw/brazil_creta.jsonl.xz filter=lfs diff=lfs merge=lfs -text
|
Multi_Legal_Pile.py
CHANGED
@@ -14,6 +14,7 @@ datasets.logging.set_verbosity_info()
|
|
14 |
logger = datasets.logging.get_logger(__name__)
|
15 |
|
16 |
_DESCRIPTION = """
|
|
|
17 |
"""
|
18 |
|
19 |
_CITATION = """
|
@@ -21,58 +22,77 @@ _CITATION = """
|
|
21 |
|
22 |
_URL = "https://huggingface.co/datasets/joelito/Multi_Legal_Pile"
|
23 |
|
24 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
_DATA_NAME = {
|
26 |
"bg": {
|
27 |
-
|
28 |
},
|
29 |
"cs": {
|
30 |
-
|
31 |
},
|
32 |
"da": {
|
33 |
-
|
34 |
-
|
35 |
},
|
36 |
"de": {
|
37 |
-
|
38 |
-
|
39 |
},
|
40 |
"en": {
|
41 |
-
|
42 |
},
|
43 |
"fr": {
|
44 |
-
|
45 |
-
|
46 |
},
|
47 |
"hu": {
|
48 |
-
|
49 |
},
|
50 |
"it": {
|
51 |
-
|
52 |
-
|
53 |
},
|
54 |
"nl": {
|
55 |
-
|
56 |
},
|
57 |
"pl": {
|
58 |
-
|
59 |
},
|
60 |
"pt": {
|
61 |
-
|
62 |
},
|
63 |
"ro": {
|
64 |
-
|
65 |
},
|
66 |
"sk": {
|
67 |
-
|
68 |
},
|
69 |
"sl": {
|
70 |
-
|
71 |
},
|
72 |
}
|
73 |
|
|
|
74 |
_PILE_OF_LAW_NAMES = {
|
75 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
"courtlistener_opinions": {
|
77 |
"train": [
|
78 |
"train.courtlisteneropinions.0",
|
@@ -85,40 +105,119 @@ _PILE_OF_LAW_NAMES = {
|
|
85 |
"train.courtlisteneropinions.7",
|
86 |
"train.courtlisteneropinions.8",
|
87 |
"train.courtlisteneropinions.9",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
],
|
89 |
"validation": [
|
90 |
"validation.courtlisteneropinions.0",
|
91 |
"validation.courtlisteneropinions.1",
|
92 |
"validation.courtlisteneropinions.2",
|
93 |
-
"validation.courtlisteneropinions.3"
|
|
|
|
|
94 |
]
|
95 |
},
|
96 |
-
"
|
97 |
-
"train": [
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
]
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
]
|
107 |
},
|
108 |
"scotus_filings": {
|
109 |
"train": ["train.scotus_docket_entries"],
|
110 |
"validation": ["validation.scotus_docket_entries"]
|
111 |
},
|
|
|
|
|
|
|
|
|
112 |
"echr": {
|
113 |
"train": ["train.echr"],
|
114 |
"validation": ["validation.echr"]
|
115 |
},
|
|
|
|
|
|
|
|
|
116 |
"canadian_decisions": {
|
117 |
"train": ["train.canadian_decisions"],
|
118 |
"validation": ["validation.canadian_decisions"]
|
119 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
120 |
},
|
121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
122 |
"atticus_contracts": {
|
123 |
"train": [
|
124 |
"train.atticus_contracts.0",
|
@@ -132,6 +231,10 @@ _PILE_OF_LAW_NAMES = {
|
|
132 |
"validation.atticus_contracts.1"
|
133 |
]
|
134 |
},
|
|
|
|
|
|
|
|
|
135 |
"edgar": {
|
136 |
"train": ["train.edgar"],
|
137 |
"validation": ["validation.edgar"]
|
@@ -140,51 +243,86 @@ _PILE_OF_LAW_NAMES = {
|
|
140 |
"train": ["train.cfpb_cc"],
|
141 |
"validation": ["validation.cfpb_cc"]
|
142 |
},
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
"train": ["train.uscode"],
|
147 |
-
"validation": ["validation.uscode"]
|
148 |
-
},
|
149 |
-
"state_codes": {
|
150 |
-
"train": ["train.state_code"],
|
151 |
-
"validation": ["validation.state_code"]
|
152 |
},
|
153 |
-
|
154 |
-
|
155 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
156 |
},
|
157 |
-
"
|
158 |
-
"train": ["train.
|
159 |
-
"validation": ["validation.
|
160 |
},
|
161 |
-
"
|
162 |
-
"train": [
|
163 |
-
|
|
|
|
|
|
|
|
|
164 |
},
|
165 |
-
"
|
166 |
-
"train": ["train.
|
167 |
-
"validation": ["validation.
|
168 |
},
|
169 |
-
},
|
170 |
-
"other": { # analyses
|
171 |
"oig": {
|
172 |
"train": ["train.oig"],
|
173 |
"validation": ["validation.oig"]
|
174 |
},
|
175 |
"olc_memos": {
|
176 |
-
"train": ["train.
|
177 |
-
"validation": ["validation.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
178 |
},
|
|
|
|
|
|
|
|
|
179 |
},
|
180 |
}
|
181 |
|
182 |
_LANGUAGES = ["bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr",
|
183 |
"hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv"]
|
184 |
|
185 |
-
_TYPES = [
|
186 |
|
187 |
-
|
188 |
"France", "Germany", "Greece", "Hungary", "Ireland", "Italy", "Latvia", "Lithuania", "Luxembourg",
|
189 |
"Malta", "Netherlands", "Poland", "Portugal", "Romania", "Slovakia", "Slovenia", "Spain", "Sweden",
|
190 |
"EU", "Switzerland", "UK", "US", "Canada", "N/A"]
|
@@ -198,7 +336,7 @@ class MultiLegalPileConfig(datasets.BuilderConfig):
|
|
198 |
Args:
|
199 |
name: combination of language and type with _
|
200 |
language: One of bg,cs,da,de,el,en,es,et,fi,fr,ga,hr,hu,it,lt,lv,mt,nl,pl,pt,ro,sk,sl,sv or all
|
201 |
-
type: One of caselaw,contracts,legislation,other or all
|
202 |
**kwargs: keyword arguments forwarded to super.
|
203 |
"""
|
204 |
super(MultiLegalPileConfig, self).__init__(**kwargs)
|
@@ -211,8 +349,11 @@ class MultiLegalPileConfig(datasets.BuilderConfig):
|
|
211 |
class MultiLegalPile(datasets.GeneratorBasedBuilder):
|
212 |
"""MultiLegalPile: A Corpus Covering Large-Scale Multilingual Legal Data"""
|
213 |
BUILDER_CONFIG_CLASS = MultiLegalPileConfig
|
|
|
214 |
|
215 |
-
BUILDER_CONFIGS = [MultiLegalPileConfig(f"{language}_{type}"
|
|
|
|
|
216 |
for type in _TYPES + ["all"]
|
217 |
for language in _LANGUAGES + ["all"]]
|
218 |
|
@@ -223,7 +364,7 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
|
|
223 |
{
|
224 |
"language": datasets.Value("string"), # one of _LANGUAGES
|
225 |
"type": datasets.Value("string"), # one of _TYPES
|
226 |
-
"jurisdiction": datasets.Value("string"), # one of
|
227 |
"text": datasets.Value("string"),
|
228 |
}
|
229 |
),
|
@@ -267,7 +408,7 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
|
|
267 |
data_infos.append(info.copy())
|
268 |
|
269 |
# add mc4_legal data
|
270 |
-
if type == '
|
271 |
info["jurisdiction"] = "N/A"
|
272 |
if language in ["de", "en", "es"]: # here we need to chunk because the files are too large
|
273 |
for idx in [0, 1]:
|
@@ -285,9 +426,9 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
|
|
285 |
for name, train_val_files in _PILE_OF_LAW_NAMES[type].items():
|
286 |
if name == 'canadian_decisions':
|
287 |
info["jurisdiction"] = "Canada"
|
288 |
-
elif name in ['echr']:
|
289 |
info["jurisdiction"] = "EU"
|
290 |
-
elif name in ['constitutions']:
|
291 |
info["jurisdiction"] = "N/A"
|
292 |
for split in ["train", "validation"]:
|
293 |
for file_name in train_val_files[split]:
|
|
|
14 |
logger = datasets.logging.get_logger(__name__)
|
15 |
|
16 |
_DESCRIPTION = """
|
17 |
+
Multi Legal Pile is a dataset of legal documents in the 24 EU languages.
|
18 |
"""
|
19 |
|
20 |
_CITATION = """
|
|
|
22 |
|
23 |
_URL = "https://huggingface.co/datasets/joelito/Multi_Legal_Pile"
|
24 |
|
25 |
+
CASELAW = "caselaw"
|
26 |
+
LEGISLATION = "legislation"
|
27 |
+
CONTRACTS = "contracts"
|
28 |
+
OTHER = "other"
|
29 |
+
MC4_LEGAL = "mc4_legal"
|
30 |
+
|
31 |
+
# Native MultiLegalPile
|
32 |
_DATA_NAME = {
|
33 |
"bg": {
|
34 |
+
LEGISLATION: ["bulgaria_marcell"],
|
35 |
},
|
36 |
"cs": {
|
37 |
+
CASELAW: ["czechia_constitutional_court", "czechia_supreme_administrative_court", "czechia_supreme_court"],
|
38 |
},
|
39 |
"da": {
|
40 |
+
CASELAW: ["denmark_ddsc"],
|
41 |
+
LEGISLATION: ["denmark_ddsc"],
|
42 |
},
|
43 |
"de": {
|
44 |
+
CASELAW: ["germany_openlegaldata", "switzerland_entscheidsuche"],
|
45 |
+
LEGISLATION: ["germany_openlegaldata", "switzerland_lexfind"],
|
46 |
},
|
47 |
"en": {
|
48 |
+
LEGISLATION: ["switzerland_lexfind", "uk_uk_lex"],
|
49 |
},
|
50 |
"fr": {
|
51 |
+
CASELAW: ["belgium_jurportal", "france_cass", "luxembourg_judoc", "switzerland_entscheidsuche"],
|
52 |
+
LEGISLATION: ["belgium_ejustice", "switzerland_lexfind"],
|
53 |
},
|
54 |
"hu": {
|
55 |
+
LEGISLATION: ["hungary_marcell"],
|
56 |
},
|
57 |
"it": {
|
58 |
+
CASELAW: ["switzerland_entscheidsuche"],
|
59 |
+
LEGISLATION: ["switzerland_lexfind"],
|
60 |
},
|
61 |
"nl": {
|
62 |
+
LEGISLATION: ["belgium_ejustice"],
|
63 |
},
|
64 |
"pl": {
|
65 |
+
LEGISLATION: ["poland_marcell"],
|
66 |
},
|
67 |
"pt": {
|
68 |
+
CASELAW: ["brazil_rulingbr", "brazil_creta"],
|
69 |
},
|
70 |
"ro": {
|
71 |
+
LEGISLATION: ["romania_marcell"],
|
72 |
},
|
73 |
"sk": {
|
74 |
+
LEGISLATION: ["slovakia_marcell"],
|
75 |
},
|
76 |
"sl": {
|
77 |
+
LEGISLATION: ["slovenia_marcell"],
|
78 |
},
|
79 |
}
|
80 |
|
81 |
+
# PileOfLaw
|
82 |
_PILE_OF_LAW_NAMES = {
|
83 |
+
CASELAW: {
|
84 |
+
"courtlistener_docket_entry_documents": {
|
85 |
+
"train": [
|
86 |
+
"train.courtlistenerdocketentries.0",
|
87 |
+
"train.courtlistenerdocketentries.1",
|
88 |
+
"train.courtlistenerdocketentries.2",
|
89 |
+
"train.courtlistenerdocketentries.3"
|
90 |
+
],
|
91 |
+
"validation": [
|
92 |
+
"validation.courtlistenerdocketentries.0",
|
93 |
+
"validation.courtlistenerdocketentries.1"
|
94 |
+
]
|
95 |
+
},
|
96 |
"courtlistener_opinions": {
|
97 |
"train": [
|
98 |
"train.courtlisteneropinions.0",
|
|
|
105 |
"train.courtlisteneropinions.7",
|
106 |
"train.courtlisteneropinions.8",
|
107 |
"train.courtlisteneropinions.9",
|
108 |
+
"train.courtlisteneropinions.10",
|
109 |
+
"train.courtlisteneropinions.11",
|
110 |
+
"train.courtlisteneropinions.12",
|
111 |
+
"train.courtlisteneropinions.13",
|
112 |
+
"train.courtlisteneropinions.14",
|
113 |
+
"train.courtlisteneropinions.15"
|
114 |
+
|
115 |
],
|
116 |
"validation": [
|
117 |
"validation.courtlisteneropinions.0",
|
118 |
"validation.courtlisteneropinions.1",
|
119 |
"validation.courtlisteneropinions.2",
|
120 |
+
"validation.courtlisteneropinions.3",
|
121 |
+
"validation.courtlisteneropinions.4",
|
122 |
+
"validation.courtlisteneropinions.5"
|
123 |
]
|
124 |
},
|
125 |
+
"bva_opinions": {
|
126 |
+
"train": ["train.bva"],
|
127 |
+
"validation": ["validation.bva"]
|
128 |
+
},
|
129 |
+
"nlrb_decisions": {
|
130 |
+
"train": ["train.nlrb_decisions"],
|
131 |
+
"validation": ["validation.nlrb_decisions"]
|
132 |
+
},
|
133 |
+
"scotus_oral_arguments": {
|
134 |
+
"train": ["train.scotus_oral"],
|
135 |
+
"validation": ["validation.scotus_oral"]
|
136 |
},
|
137 |
"scotus_filings": {
|
138 |
"train": ["train.scotus_docket_entries"],
|
139 |
"validation": ["validation.scotus_docket_entries"]
|
140 |
},
|
141 |
+
"ftc_advisory_opinions": {
|
142 |
+
"train": ["train.ftc_advisory_opinions"],
|
143 |
+
"validation": ["validation.ftc_advisory_opinions"]
|
144 |
+
},
|
145 |
"echr": {
|
146 |
"train": ["train.echr"],
|
147 |
"validation": ["validation.echr"]
|
148 |
},
|
149 |
+
"tax_rulings": {
|
150 |
+
"train": ["train.taxrulings"],
|
151 |
+
"validation": ["validation.taxrulings"]
|
152 |
+
},
|
153 |
"canadian_decisions": {
|
154 |
"train": ["train.canadian_decisions"],
|
155 |
"validation": ["validation.canadian_decisions"]
|
156 |
},
|
157 |
+
"eoir": {
|
158 |
+
"train": ["train.eoir"],
|
159 |
+
"validation": ["validation.eoir"]
|
160 |
+
},
|
161 |
+
"dol_ecab": {
|
162 |
+
"train": ["train.dol_ecab"],
|
163 |
+
"validation": ["validation.dol_ecab"]
|
164 |
+
},
|
165 |
+
"icj-pcij": {
|
166 |
+
"train": ["train.ijc"],
|
167 |
+
"validation": ["validation.ijc"]
|
168 |
+
},
|
169 |
+
"hhs_alj_opinions": {
|
170 |
+
"train": ["train.hhs_alj"],
|
171 |
+
"validation": ["validation.hhs_alj"]
|
172 |
+
},
|
173 |
+
"sec_administrative_proceedings": {
|
174 |
+
"train": ["train.sec"],
|
175 |
+
"validation": ["validation.sec"]
|
176 |
+
},
|
177 |
},
|
178 |
+
LEGISLATION: {
|
179 |
+
"federal_register": {
|
180 |
+
"train": ["train.federal_register"],
|
181 |
+
"validation": ["validation.federal_register"]
|
182 |
+
},
|
183 |
+
"us_bills": {
|
184 |
+
"train": ["train.us_bills"],
|
185 |
+
"validation": ["validation.us_bills"]
|
186 |
+
},
|
187 |
+
"cfr": {
|
188 |
+
"train": ["train.cfr"],
|
189 |
+
"validation": ["validation.cfr"]
|
190 |
+
},
|
191 |
+
"state_codes": {
|
192 |
+
"train": ["train.state_code"],
|
193 |
+
"validation": ["validation.state_code"]
|
194 |
+
},
|
195 |
+
"constitutions": {
|
196 |
+
"train": ["train.constitutions"],
|
197 |
+
"validation": ["validation.constitutions"]
|
198 |
+
},
|
199 |
+
"uscode": {
|
200 |
+
"train": ["train.uscode"],
|
201 |
+
"validation": ["validation.uscode"]
|
202 |
+
},
|
203 |
+
"founding_docs": {
|
204 |
+
"train": ["train.founding_docs"],
|
205 |
+
"validation": ["validation.founding_docs"]
|
206 |
+
},
|
207 |
+
"eurlex": {
|
208 |
+
"train": ["train.eurlex"],
|
209 |
+
"validation": ["validation.eurlex"]
|
210 |
+
},
|
211 |
+
"fre": {
|
212 |
+
"train": ["train.fre"],
|
213 |
+
"validation": ["validation.fre"]
|
214 |
+
},
|
215 |
+
"frcp": {
|
216 |
+
"train": ["train.frcp"],
|
217 |
+
"validation": ["validation.frcp"]
|
218 |
+
},
|
219 |
+
},
|
220 |
+
CONTRACTS: {
|
221 |
"atticus_contracts": {
|
222 |
"train": [
|
223 |
"train.atticus_contracts.0",
|
|
|
231 |
"validation.atticus_contracts.1"
|
232 |
]
|
233 |
},
|
234 |
+
"tos": {
|
235 |
+
"train": ["train.tos"],
|
236 |
+
"validation": ["validation.tos"]
|
237 |
+
},
|
238 |
"edgar": {
|
239 |
"train": ["train.edgar"],
|
240 |
"validation": ["validation.edgar"]
|
|
|
243 |
"train": ["train.cfpb_cc"],
|
244 |
"validation": ["validation.cfpb_cc"]
|
245 |
},
|
246 |
+
"resource_contracts": {
|
247 |
+
"train": ["train.resource_contracts"],
|
248 |
+
"validation": ["validation.resource_contracts"]
|
|
|
|
|
|
|
|
|
|
|
|
|
249 |
},
|
250 |
+
},
|
251 |
+
OTHER: {
|
252 |
+
"r_legaladvice":
|
253 |
+
{
|
254 |
+
"train": ["train.r_legaldvice"],
|
255 |
+
"validation": ["validation.r_legaldvice"]
|
256 |
+
},
|
257 |
+
"cc_casebooks": {
|
258 |
+
"train": ["train.cc_casebooks"],
|
259 |
+
"validation": ["validation.cc_casebooks"]
|
260 |
},
|
261 |
+
"euro_parl": {
|
262 |
+
"train": ["train.euro_parl"],
|
263 |
+
"validation": ["validation.euro_parl"]
|
264 |
},
|
265 |
+
"exam_outlines": {
|
266 |
+
"train": [
|
267 |
+
"train.examoutlines",
|
268 |
+
],
|
269 |
+
"validation": [
|
270 |
+
"validation.examoutlines",
|
271 |
+
]
|
272 |
},
|
273 |
+
"congressional_hearings": {
|
274 |
+
"train": ["train.congressional_hearings"],
|
275 |
+
"validation": ["validation.congressional_hearings"]
|
276 |
},
|
|
|
|
|
277 |
"oig": {
|
278 |
"train": ["train.oig"],
|
279 |
"validation": ["validation.oig"]
|
280 |
},
|
281 |
"olc_memos": {
|
282 |
+
"train": ["train.olc_memos"],
|
283 |
+
"validation": ["validation.olc_memos"]
|
284 |
+
},
|
285 |
+
"un_debates": {
|
286 |
+
"train": ["train.undebates"],
|
287 |
+
"validation": ["validation.undebates"]
|
288 |
+
},
|
289 |
+
"uspto_office_actions": {
|
290 |
+
"train": ["train.uspto_oab"],
|
291 |
+
"validation": ["validation.uspto_oab"]
|
292 |
+
},
|
293 |
+
"ed_policy_guidance": {
|
294 |
+
"train": ["train.ed_policy_guidance"],
|
295 |
+
"validation": ["validation.ed_policy_guidance"]
|
296 |
+
},
|
297 |
+
"acus_reports": {
|
298 |
+
"train": ["train.acus_reports"],
|
299 |
+
"validation": ["validation.acus_reports"]
|
300 |
+
},
|
301 |
+
"fmshrc_bluebooks": {
|
302 |
+
"train": ["train.fmshrc"],
|
303 |
+
"validation": ["validation.fmshrc"]
|
304 |
+
},
|
305 |
+
"medicaid_policy_guidance": {
|
306 |
+
"train": ["train.medicaid_policy_guidance"],
|
307 |
+
"validation": ["validation.medicaid_policy_guidance"]
|
308 |
+
},
|
309 |
+
"irs_legal_advice_memos": {
|
310 |
+
"train": ["train.irs_legal_advice_memos"],
|
311 |
+
"validation": ["validation.irs_legal_advice_memos"]
|
312 |
},
|
313 |
+
"doj_guidance_documents": {
|
314 |
+
"train": ["train.doj_guidance"],
|
315 |
+
"validation": ["validation.doj_guidance"]
|
316 |
+
}
|
317 |
},
|
318 |
}
|
319 |
|
320 |
_LANGUAGES = ["bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr",
|
321 |
"hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv"]
|
322 |
|
323 |
+
_TYPES = [CASELAW, CONTRACTS, LEGISLATION, OTHER, MC4_LEGAL]
|
324 |
|
325 |
+
_JURISDICTIONS = ["Austria", "Belgium", "Brazil", "Bulgaria", "Croatia", "Czechia", "Denmark", "Estonia", "Finland",
|
326 |
"France", "Germany", "Greece", "Hungary", "Ireland", "Italy", "Latvia", "Lithuania", "Luxembourg",
|
327 |
"Malta", "Netherlands", "Poland", "Portugal", "Romania", "Slovakia", "Slovenia", "Spain", "Sweden",
|
328 |
"EU", "Switzerland", "UK", "US", "Canada", "N/A"]
|
|
|
336 |
Args:
|
337 |
name: combination of language and type with _
|
338 |
language: One of bg,cs,da,de,el,en,es,et,fi,fr,ga,hr,hu,it,lt,lv,mt,nl,pl,pt,ro,sk,sl,sv or all
|
339 |
+
type: One of caselaw,contracts,legislation,other,mc4_legal or all
|
340 |
**kwargs: keyword arguments forwarded to super.
|
341 |
"""
|
342 |
super(MultiLegalPileConfig, self).__init__(**kwargs)
|
|
|
349 |
class MultiLegalPile(datasets.GeneratorBasedBuilder):
|
350 |
"""MultiLegalPile: A Corpus Covering Large-Scale Multilingual Legal Data"""
|
351 |
BUILDER_CONFIG_CLASS = MultiLegalPileConfig
|
352 |
+
VERSION = datasets.Version("2.0.0", "")
|
353 |
|
354 |
+
BUILDER_CONFIGS = [MultiLegalPileConfig(f"{language}_{type}",
|
355 |
+
version=VERSION,
|
356 |
+
description=f"MultiLegalPile for language {language} and type {type}")
|
357 |
for type in _TYPES + ["all"]
|
358 |
for language in _LANGUAGES + ["all"]]
|
359 |
|
|
|
364 |
{
|
365 |
"language": datasets.Value("string"), # one of _LANGUAGES
|
366 |
"type": datasets.Value("string"), # one of _TYPES
|
367 |
+
"jurisdiction": datasets.Value("string"), # one of _JURISDICTIONS
|
368 |
"text": datasets.Value("string"),
|
369 |
}
|
370 |
),
|
|
|
408 |
data_infos.append(info.copy())
|
409 |
|
410 |
# add mc4_legal data
|
411 |
+
if type == 'mc4_legal':
|
412 |
info["jurisdiction"] = "N/A"
|
413 |
if language in ["de", "en", "es"]: # here we need to chunk because the files are too large
|
414 |
for idx in [0, 1]:
|
|
|
426 |
for name, train_val_files in _PILE_OF_LAW_NAMES[type].items():
|
427 |
if name == 'canadian_decisions':
|
428 |
info["jurisdiction"] = "Canada"
|
429 |
+
elif name in ['echr', 'eurlex', 'euro_parl']:
|
430 |
info["jurisdiction"] = "EU"
|
431 |
+
elif name in ['constitutions', 'icj-pcij', 'r_legaladvice', 'un_debates', 'tos']:
|
432 |
info["jurisdiction"] = "N/A"
|
433 |
for split in ["train", "validation"]:
|
434 |
for file_name in train_val_files[split]:
|
README.md
CHANGED
@@ -1,44 +1,45 @@
|
|
1 |
---
|
2 |
annotations_creators:
|
|
|
3 |
- other
|
4 |
-
language_creators:
|
5 |
- found
|
6 |
-
language:
|
7 |
-
- bg
|
8 |
-
- cs
|
9 |
-
- da
|
10 |
-
- de
|
11 |
-
- el
|
12 |
- en
|
13 |
-
- es
|
14 |
-
- et
|
15 |
-
- fi
|
16 |
-
- fr
|
17 |
- ga
|
18 |
-
- hr
|
19 |
-
- hu
|
20 |
-
- it
|
21 |
-
- lt
|
22 |
-
- lv
|
23 |
- mt
|
24 |
-
- nl
|
25 |
-
- pl
|
26 |
-
- pt
|
27 |
-
- ro
|
28 |
-
- sk
|
29 |
-
- sl
|
30 |
- sv
|
31 |
-
license:
|
32 |
- cc-by-4.0
|
33 |
-
multilinguality:
|
34 |
- multilingual
|
35 |
-
paperswithcode_id: null
|
36 |
-
pretty_name: "MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain"
|
37 |
-
size_categories:
|
38 |
- 10M<n<100M
|
39 |
-
source_datasets:
|
40 |
- original
|
41 |
-
task_categories:
|
42 |
- fill-mask
|
43 |
|
44 |
---
|
@@ -74,15 +75,15 @@ task_categories:
|
|
74 |
## Dataset Description
|
75 |
|
76 |
- **Homepage:**
|
77 |
-
- **Repository:**
|
78 |
-
- **Paper:**
|
79 |
- **Leaderboard:**
|
80 |
- **Point of Contact:** [Joel Niklaus](mailto:[email protected])
|
81 |
|
82 |
### Dataset Summary
|
83 |
|
84 |
The Multi_Legal_Pile is a large-scale multilingual legal dataset suited for pretraining language models.
|
85 |
-
It spans over 24 languages and
|
86 |
|
87 |
### Supported Tasks and Leaderboards
|
88 |
|
@@ -96,21 +97,25 @@ ro, sk, sl, sv
|
|
96 |
## Dataset Structure
|
97 |
|
98 |
It is structured in the following format:
|
99 |
-
|
100 |
|
101 |
-
|
102 |
|
103 |
- caselaw
|
104 |
- contracts
|
105 |
- legislation
|
106 |
- other
|
|
|
107 |
|
|
|
|
|
108 |
|
109 |
Use the dataset like this:
|
|
|
110 |
```python
|
111 |
from datasets import load_dataset
|
112 |
|
113 |
-
config = 'en_contracts'
|
114 |
dataset = load_dataset('joelito/Multi_Legal_Pile', config, split='train', streaming=True)
|
115 |
```
|
116 |
|
@@ -122,17 +127,16 @@ all_legislation').
|
|
122 |
|
123 |
The file format is jsonl.xz and there is one split available ("train").
|
124 |
|
|
|
125 |
|
126 |
-
|
127 |
-
- Native Multi Legal Pile (30GB)
|
128 |
- Eurlex Resources (179GB)
|
129 |
- MC4 Legal (133GB)
|
130 |
-
- Pile of Law (
|
131 |
-
|
132 |
|
133 |
#### Native Multilingual Legal Pile data
|
134 |
|
135 |
-
| Language | Text Type |
|
136 |
|:---------|:------------|:-------------|:-----------------------------------|----------:|-------:|----------:|---------------:|-----|--------:|
|
137 |
| bg | legislation | Bulgaria | MARCELL | 588 | xxx | xxx | xxx | | |
|
138 |
| cs | caselaw | Czechia | CzCDC Constitutional Court | 713 | xxx | xxx | xxx | | |
|
@@ -158,10 +162,11 @@ The complete dataset (564GB) consists of four large subsets:
|
|
158 |
| nl | legislation | Belgium | ejustice | 178 | xxx | xxx | xxx | | |
|
159 |
| pl | legislation | Poland | MARCELL | 264 | xxx | xxx | xxx | | |
|
160 |
| pt | caselaw | Brazil | RulingBR | 173 | xxx | xxx | xxx | | |
|
|
|
161 |
| ro | legislation | Romania | MARCELL | 2704 | xxx | xxx | xxx | | |
|
162 |
| sk | legislation | Slovakia | MARCELL | 192 | xxx | xxx | xxx | | |
|
163 |
| sl | legislation | Slovenia | MARCELL | 753 | xxx | xxx | xxx | | |
|
164 |
-
| total | all | all | all |
|
165 |
|
166 |
#### Eurlex Resources
|
167 |
|
@@ -175,15 +180,6 @@ See [MC4 Legal](https://huggingface.co/datasets/joelito/mc4_legal#data-instances
|
|
175 |
|
176 |
See [Pile-of-Law](https://huggingface.co/datasets/pile-of-law/pile-of-law#data-instances) for more information.
|
177 |
|
178 |
-
For simplicity and with respect to balancing off data across different jurisdictions and languages,
|
179 |
-
we disregard many US resources that are either very specialized (e.g., tax rulings),
|
180 |
-
outdated/historical (e.g., founding letters),
|
181 |
-
very small (less than 20MB),
|
182 |
-
not legal language in the strict sense (conversations),
|
183 |
-
or the information overlaps with other sources (e.g. study materials or eurlex).
|
184 |
-
If you are interested in a US-based (US-biased) model, refer to the "Pile of (US) Law" by Henderson et al. (2022).
|
185 |
-
Analyses are put into the "other" category because in mc4_legal we also likely have similar text.
|
186 |
-
|
187 |
| Language | Type | Jurisdiction | Source | Size (MB) | Tokens | Documents | Tokens/Document | Part of Multi_Legal_Pile |
|
188 |
|:-----------|:------------|:---------------|:-------------------------------------|------------:|------------:|------------:|------------------:|:---------------------------|
|
189 |
| en | all | all | all | 503712 | 50547777921 | 9872444 | 5120 | yes |
|
@@ -246,7 +242,6 @@ Analyses are put into the "other" category because in mc4_legal we also likely h
|
|
246 |
|
247 |
[More Information Needed]
|
248 |
|
249 |
-
|
250 |
### Annotations
|
251 |
|
252 |
#### Annotation process
|
|
|
1 |
---
|
2 |
annotations_creators:
|
3 |
+
|
4 |
- other
|
5 |
+
language_creators:
|
6 |
- found
|
7 |
+
language:
|
8 |
+
- bg
|
9 |
+
- cs
|
10 |
+
- da
|
11 |
+
- de
|
12 |
+
- el
|
13 |
- en
|
14 |
+
- es
|
15 |
+
- et
|
16 |
+
- fi
|
17 |
+
- fr
|
18 |
- ga
|
19 |
+
- hr
|
20 |
+
- hu
|
21 |
+
- it
|
22 |
+
- lt
|
23 |
+
- lv
|
24 |
- mt
|
25 |
+
- nl
|
26 |
+
- pl
|
27 |
+
- pt
|
28 |
+
- ro
|
29 |
+
- sk
|
30 |
+
- sl
|
31 |
- sv
|
32 |
+
license:
|
33 |
- cc-by-4.0
|
34 |
+
multilinguality:
|
35 |
- multilingual
|
36 |
+
paperswithcode_id: null
|
37 |
+
pretty_name: "MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain"
|
38 |
+
size_categories:
|
39 |
- 10M<n<100M
|
40 |
+
source_datasets:
|
41 |
- original
|
42 |
+
task_categories:
|
43 |
- fill-mask
|
44 |
|
45 |
---
|
|
|
75 |
## Dataset Description
|
76 |
|
77 |
- **Homepage:**
|
78 |
+
- **Repository:**
|
79 |
+
- **Paper:**
|
80 |
- **Leaderboard:**
|
81 |
- **Point of Contact:** [Joel Niklaus](mailto:[email protected])
|
82 |
|
83 |
### Dataset Summary
|
84 |
|
85 |
The Multi_Legal_Pile is a large-scale multilingual legal dataset suited for pretraining language models.
|
86 |
+
It spans over 24 languages and five legal text types.
|
87 |
|
88 |
### Supported Tasks and Leaderboards
|
89 |
|
|
|
97 |
## Dataset Structure
|
98 |
|
99 |
It is structured in the following format:
|
100 |
+
type -> language -> jurisdiction.jsonl.xz
|
101 |
|
102 |
+
type is one of the following:
|
103 |
|
104 |
- caselaw
|
105 |
- contracts
|
106 |
- legislation
|
107 |
- other
|
108 |
+
- mc4_legal
|
109 |
|
110 |
+
`mc4_legal` is a subset of the other type but is listed separately so it can be easily excluded since it is less
|
111 |
+
permissively licensed than the other types.
|
112 |
|
113 |
Use the dataset like this:
|
114 |
+
|
115 |
```python
|
116 |
from datasets import load_dataset
|
117 |
|
118 |
+
config = 'en_contracts' # {language}_{type}
|
119 |
dataset = load_dataset('joelito/Multi_Legal_Pile', config, split='train', streaming=True)
|
120 |
```
|
121 |
|
|
|
127 |
|
128 |
The file format is jsonl.xz and there is one split available ("train").
|
129 |
|
130 |
+
The complete dataset (653GB) consists of four large subsets:
|
131 |
|
132 |
+
- Native Multi Legal Pile (49GB)
|
|
|
133 |
- Eurlex Resources (179GB)
|
134 |
- MC4 Legal (133GB)
|
135 |
+
- Pile of Law (292GB)
|
|
|
136 |
|
137 |
#### Native Multilingual Legal Pile data
|
138 |
|
139 |
+
| Language | Text Type | Jurisdiction | Source | Size (MB) | Tokens | Documents | Words/Document | URL | License |
|
140 |
|:---------|:------------|:-------------|:-----------------------------------|----------:|-------:|----------:|---------------:|-----|--------:|
|
141 |
| bg | legislation | Bulgaria | MARCELL | 588 | xxx | xxx | xxx | | |
|
142 |
| cs | caselaw | Czechia | CzCDC Constitutional Court | 713 | xxx | xxx | xxx | | |
|
|
|
162 |
| nl | legislation | Belgium | ejustice | 178 | xxx | xxx | xxx | | |
|
163 |
| pl | legislation | Poland | MARCELL | 264 | xxx | xxx | xxx | | |
|
164 |
| pt | caselaw | Brazil | RulingBR | 173 | xxx | xxx | xxx | | |
|
165 |
+
| pt | caselaw | Brazil | CRETA | 19000 | xxx | xxx | xxx | | |
|
166 |
| ro | legislation | Romania | MARCELL | 2704 | xxx | xxx | xxx | | |
|
167 |
| sk | legislation | Slovakia | MARCELL | 192 | xxx | xxx | xxx | | |
|
168 |
| sl | legislation | Slovenia | MARCELL | 753 | xxx | xxx | xxx | | |
|
169 |
+
| total | all | all | all | 48759 | xxx | xxx | xxx | | |
|
170 |
|
171 |
#### Eurlex Resources
|
172 |
|
|
|
180 |
|
181 |
See [Pile-of-Law](https://huggingface.co/datasets/pile-of-law/pile-of-law#data-instances) for more information.
|
182 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
183 |
| Language | Type | Jurisdiction | Source | Size (MB) | Tokens | Documents | Tokens/Document | Part of Multi_Legal_Pile |
|
184 |
|:-----------|:------------|:---------------|:-------------------------------------|------------:|------------:|------------:|------------------:|:---------------------------|
|
185 |
| en | all | all | all | 503712 | 50547777921 | 9872444 | 5120 | yes |
|
|
|
242 |
|
243 |
[More Information Needed]
|
244 |
|
|
|
245 |
### Annotations
|
246 |
|
247 |
#### Annotation process
|