updated data loader to include update from pile of law

.gitattributes

@@ -77,3 +77,4 @@ data/fr/caselaw/switzerland_entscheidsuche.jsonl.xz filter=lfs diff=lfs merge=lf
 data/it/caselaw/switzerland_entscheidsuche.jsonl.xz filter=lfs diff=lfs merge=lfs -text
 data/en/legislation/uk_uk_lex.jsonl.xz filter=lfs diff=lfs merge=lfs -text
 data/hu/legislation/hungary_marcell.jsonl.xz filter=lfs diff=lfs merge=lfs -text
+data/pt/caselaw/brazil_creta.jsonl.xz filter=lfs diff=lfs merge=lfs -text

Multi_Legal_Pile.py

@@ -14,6 +14,7 @@ datasets.logging.set_verbosity_info()
 logger = datasets.logging.get_logger(__name__)
 
 _DESCRIPTION = """
+Multi Legal Pile is a dataset of legal documents in the 24 EU languages.
 """
 
 _CITATION = """
@@ -21,58 +22,77 @@ _CITATION = """
 
 _URL = "https://huggingface.co/datasets/joelito/Multi_Legal_Pile"
 
-# Specific Data
+CASELAW = "caselaw"
+LEGISLATION = "legislation"
+CONTRACTS = "contracts"
+OTHER = "other"
+MC4_LEGAL = "mc4_legal"
+
+# Native MultiLegalPile
 _DATA_NAME = {
     "bg": {
-        "legislation": ["bulgaria_marcell"],
+        LEGISLATION: ["bulgaria_marcell"],
     },
     "cs": {
-        "caselaw": ["czechia_constitutional_court", "czechia_supreme_administrative_court", "czechia_supreme_court"],
+        CASELAW: ["czechia_constitutional_court", "czechia_supreme_administrative_court", "czechia_supreme_court"],
     },
     "da": {
-        "caselaw": ["denmark_ddsc"],
-        "legislation": ["denmark_ddsc"],
+        CASELAW: ["denmark_ddsc"],
+        LEGISLATION: ["denmark_ddsc"],
     },
     "de": {
-        "caselaw": ["germany_openlegaldata", "switzerland_entscheidsuche"],
-        "legislation": ["germany_openlegaldata", "switzerland_lexfind"],
+        CASELAW: ["germany_openlegaldata", "switzerland_entscheidsuche"],
+        LEGISLATION: ["germany_openlegaldata", "switzerland_lexfind"],
     },
     "en": {
-        "legislation": ["switzerland_lexfind", "uk_uk_lex"],
+        LEGISLATION: ["switzerland_lexfind", "uk_uk_lex"],
     },
     "fr": {
-        "caselaw": ["belgium_jurportal", "france_cass", "luxembourg_judoc", "switzerland_entscheidsuche"],
-        "legislation": ["belgium_ejustice", "switzerland_lexfind"],
+        CASELAW: ["belgium_jurportal", "france_cass", "luxembourg_judoc", "switzerland_entscheidsuche"],
+        LEGISLATION: ["belgium_ejustice", "switzerland_lexfind"],
     },
     "hu": {
-        "legislation": ["hungary_marcell"],
+        LEGISLATION: ["hungary_marcell"],
     },
     "it": {
-        "caselaw": ["switzerland_entscheidsuche"],
-        "legislation": ["switzerland_lexfind"],
+        CASELAW: ["switzerland_entscheidsuche"],
+        LEGISLATION: ["switzerland_lexfind"],
     },
     "nl": {
-        "legislation": ["belgium_ejustice"],
+        LEGISLATION: ["belgium_ejustice"],
     },
     "pl": {
-        "legislation": ["poland_marcell"],
+        LEGISLATION: ["poland_marcell"],
     },
     "pt": {
-        "caselaw": ["brazil_rulingbr"],
+        CASELAW: ["brazil_rulingbr", "brazil_creta"],
     },
     "ro": {
-        "legislation": ["romania_marcell"],
+        LEGISLATION: ["romania_marcell"],
     },
     "sk": {
-        "legislation": ["slovakia_marcell"],
+        LEGISLATION: ["slovakia_marcell"],
     },
     "sl": {
-        "legislation": ["slovenia_marcell"],
+        LEGISLATION: ["slovenia_marcell"],
     },
 }
 
+# PileOfLaw
 _PILE_OF_LAW_NAMES = {
-    "caselaw": {
+    CASELAW: {
+        "courtlistener_docket_entry_documents": {
+            "train": [
+                "train.courtlistenerdocketentries.0",
+                "train.courtlistenerdocketentries.1",
+                "train.courtlistenerdocketentries.2",
+                "train.courtlistenerdocketentries.3"
+            ],
+            "validation": [
+                "validation.courtlistenerdocketentries.0",
+                "validation.courtlistenerdocketentries.1"
+            ]
+        },
         "courtlistener_opinions": {
             "train": [
                 "train.courtlisteneropinions.0",
@@ -85,40 +105,119 @@ _PILE_OF_LAW_NAMES = {
                 "train.courtlisteneropinions.7",
                 "train.courtlisteneropinions.8",
                 "train.courtlisteneropinions.9",
+                "train.courtlisteneropinions.10",
+                "train.courtlisteneropinions.11",
+                "train.courtlisteneropinions.12",
+                "train.courtlisteneropinions.13",
+                "train.courtlisteneropinions.14",
+                "train.courtlisteneropinions.15"
+
             ],
             "validation": [
                 "validation.courtlisteneropinions.0",
                 "validation.courtlisteneropinions.1",
                 "validation.courtlisteneropinions.2",
-                "validation.courtlisteneropinions.3"
+                "validation.courtlisteneropinions.3",
+                "validation.courtlisteneropinions.4",
+                "validation.courtlisteneropinions.5"
             ]
         },
-        "courtlistener_docket_entry_documents": {
-            "train": [
-                "train.courtlistenerdocketentries.0",
-                "train.courtlistenerdocketentries.1",
-                "train.courtlistenerdocketentries.2",
-                "train.courtlistenerdocketentries.3"
-            ],
-            "validation": [
-                "validation.courtlistenerdocketentries.0",
-                "validation.courtlistenerdocketentries.1"
-            ]
+        "bva_opinions": {
+            "train": ["train.bva"],
+            "validation": ["validation.bva"]
+        },
+        "nlrb_decisions": {
+            "train": ["train.nlrb_decisions"],
+            "validation": ["validation.nlrb_decisions"]
+        },
+        "scotus_oral_arguments": {
+            "train": ["train.scotus_oral"],
+            "validation": ["validation.scotus_oral"]
         },
         "scotus_filings": {
             "train": ["train.scotus_docket_entries"],
             "validation": ["validation.scotus_docket_entries"]
         },
+        "ftc_advisory_opinions": {
+            "train": ["train.ftc_advisory_opinions"],
+            "validation": ["validation.ftc_advisory_opinions"]
+        },
         "echr": {
             "train": ["train.echr"],
             "validation": ["validation.echr"]
         },
+        "tax_rulings": {
+            "train": ["train.taxrulings"],
+            "validation": ["validation.taxrulings"]
+        },
         "canadian_decisions": {
             "train": ["train.canadian_decisions"],
             "validation": ["validation.canadian_decisions"]
         },
+        "eoir": {
+            "train": ["train.eoir"],
+            "validation": ["validation.eoir"]
+        },
+        "dol_ecab": {
+            "train": ["train.dol_ecab"],
+            "validation": ["validation.dol_ecab"]
+        },
+        "icj-pcij": {
+            "train": ["train.ijc"],
+            "validation": ["validation.ijc"]
+        },
+        "hhs_alj_opinions": {
+            "train": ["train.hhs_alj"],
+            "validation": ["validation.hhs_alj"]
+        },
+        "sec_administrative_proceedings": {
+            "train": ["train.sec"],
+            "validation": ["validation.sec"]
+        },
     },
-    "contracts": {
+    LEGISLATION: {
+        "federal_register": {
+            "train": ["train.federal_register"],
+            "validation": ["validation.federal_register"]
+        },
+        "us_bills": {
+            "train": ["train.us_bills"],
+            "validation": ["validation.us_bills"]
+        },
+        "cfr": {
+            "train": ["train.cfr"],
+            "validation": ["validation.cfr"]
+        },
+        "state_codes": {
+            "train": ["train.state_code"],
+            "validation": ["validation.state_code"]
+        },
+        "constitutions": {
+            "train": ["train.constitutions"],
+            "validation": ["validation.constitutions"]
+        },
+        "uscode": {
+            "train": ["train.uscode"],
+            "validation": ["validation.uscode"]
+        },
+        "founding_docs": {
+            "train": ["train.founding_docs"],
+            "validation": ["validation.founding_docs"]
+        },
+        "eurlex": {
+            "train": ["train.eurlex"],
+            "validation": ["validation.eurlex"]
+        },
+        "fre": {
+            "train": ["train.fre"],
+            "validation": ["validation.fre"]
+        },
+        "frcp": {
+            "train": ["train.frcp"],
+            "validation": ["validation.frcp"]
+        },
+    },
+    CONTRACTS: {
         "atticus_contracts": {
             "train": [
                 "train.atticus_contracts.0",
@@ -132,6 +231,10 @@ _PILE_OF_LAW_NAMES = {
                 "validation.atticus_contracts.1"
             ]
         },
+        "tos": {
+            "train": ["train.tos"],
+            "validation": ["validation.tos"]
+        },
         "edgar": {
             "train": ["train.edgar"],
             "validation": ["validation.edgar"]
@@ -140,51 +243,86 @@ _PILE_OF_LAW_NAMES = {
             "train": ["train.cfpb_cc"],
             "validation": ["validation.cfpb_cc"]
         },
-    },
-    "legislation": {
-        "uscode": {
-            "train": ["train.uscode"],
-            "validation": ["validation.uscode"]
-        },
-        "state_codes": {
-            "train": ["train.state_code"],
-            "validation": ["validation.state_code"]
+        "resource_contracts": {
+            "train": ["train.resource_contracts"],
+            "validation": ["validation.resource_contracts"]
         },
-        "us_bills": {
-            "train": ["train.us_bills"],
-            "validation": ["validation.us_bills"]
+    },
+    OTHER: {
+        "r_legaladvice":
+            {
+                "train": ["train.r_legaldvice"],
+                "validation": ["validation.r_legaldvice"]
+            },
+        "cc_casebooks": {
+            "train": ["train.cc_casebooks"],
+            "validation": ["validation.cc_casebooks"]
         },
-        "federal_register": {
-            "train": ["train.federal_register"],
-            "validation": ["validation.federal_register"]
+        "euro_parl": {
+            "train": ["train.euro_parl"],
+            "validation": ["validation.euro_parl"]
         },
-        "cfr": {
-            "train": ["train.cfr"],
-            "validation": ["validation.cfr"]
+        "exam_outlines": {
+            "train": [
+                "train.examoutlines",
+            ],
+            "validation": [
+                "validation.examoutlines",
+            ]
         },
-        "constitutions": {
-            "train": ["train.constitutions"],
-            "validation": ["validation.constitutions"]
+        "congressional_hearings": {
+            "train": ["train.congressional_hearings"],
+            "validation": ["validation.congressional_hearings"]
         },
-    },
-    "other": {  # analyses
         "oig": {
             "train": ["train.oig"],
             "validation": ["validation.oig"]
         },
         "olc_memos": {
-            "train": ["train.olcmemos"],
-            "validation": ["validation.olcmemos"]
+            "train": ["train.olc_memos"],
+            "validation": ["validation.olc_memos"]
+        },
+        "un_debates": {
+            "train": ["train.undebates"],
+            "validation": ["validation.undebates"]
+        },
+        "uspto_office_actions": {
+            "train": ["train.uspto_oab"],
+            "validation": ["validation.uspto_oab"]
+        },
+        "ed_policy_guidance": {
+            "train": ["train.ed_policy_guidance"],
+            "validation": ["validation.ed_policy_guidance"]
+        },
+        "acus_reports": {
+            "train": ["train.acus_reports"],
+            "validation": ["validation.acus_reports"]
+        },
+        "fmshrc_bluebooks": {
+            "train": ["train.fmshrc"],
+            "validation": ["validation.fmshrc"]
+        },
+        "medicaid_policy_guidance": {
+            "train": ["train.medicaid_policy_guidance"],
+            "validation": ["validation.medicaid_policy_guidance"]
+        },
+        "irs_legal_advice_memos": {
+            "train": ["train.irs_legal_advice_memos"],
+            "validation": ["validation.irs_legal_advice_memos"]
         },
+        "doj_guidance_documents": {
+            "train": ["train.doj_guidance"],
+            "validation": ["validation.doj_guidance"]
+        }
     },
 }
 
 _LANGUAGES = ["bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr",
               "hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv"]
 
-_TYPES = ["caselaw", "contracts", "legislation", "other"]
+_TYPES = [CASELAW, CONTRACTS, LEGISLATION, OTHER, MC4_LEGAL]
 
-_JURISDICTONS = ["Austria", "Belgium", "Bulgaria", "Croatia", "Czechia", "Denmark", "Estonia", "Finland",
+_JURISDICTIONS = ["Austria", "Belgium", "Brazil", "Bulgaria", "Croatia", "Czechia", "Denmark", "Estonia", "Finland",
                  "France", "Germany", "Greece", "Hungary", "Ireland", "Italy", "Latvia", "Lithuania", "Luxembourg",
                  "Malta", "Netherlands", "Poland", "Portugal", "Romania", "Slovakia", "Slovenia", "Spain", "Sweden",
                  "EU", "Switzerland", "UK", "US", "Canada", "N/A"]
@@ -198,7 +336,7 @@ class MultiLegalPileConfig(datasets.BuilderConfig):
         Args:
             name: combination of language and type with _
             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
-            type:           One of caselaw,contracts,legislation,other or all
+            type:           One of caselaw,contracts,legislation,other,mc4_legal or all
           **kwargs: keyword arguments forwarded to super.
         """
         super(MultiLegalPileConfig, self).__init__(**kwargs)
@@ -211,8 +349,11 @@ class MultiLegalPileConfig(datasets.BuilderConfig):
 class MultiLegalPile(datasets.GeneratorBasedBuilder):
     """MultiLegalPile: A Corpus Covering Large-Scale Multilingual Legal Data"""
     BUILDER_CONFIG_CLASS = MultiLegalPileConfig
+    VERSION = datasets.Version("2.0.0", "")
 
-    BUILDER_CONFIGS = [MultiLegalPileConfig(f"{language}_{type}")
+    BUILDER_CONFIGS = [MultiLegalPileConfig(f"{language}_{type}",
+                                            version=VERSION,
+                                            description=f"MultiLegalPile for language {language} and type {type}")
                        for type in _TYPES + ["all"]
                        for language in _LANGUAGES + ["all"]]
 
@@ -223,7 +364,7 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
                 {
                     "language": datasets.Value("string"),  # one of _LANGUAGES
                     "type": datasets.Value("string"),  # one of _TYPES
-                    "jurisdiction": datasets.Value("string"),  # one of _JURISDICTONS
+                    "jurisdiction": datasets.Value("string"),  # one of _JURISDICTIONS
                     "text": datasets.Value("string"),
                 }
             ),
@@ -267,7 +408,7 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
                         data_infos.append(info.copy())
 
                     # add mc4_legal data
-                    if type == 'other':
+                    if type == 'mc4_legal':
                         info["jurisdiction"] = "N/A"
                         if language in ["de", "en", "es"]:  # here we need to chunk because the files are too large
                             for idx in [0, 1]:
@@ -285,9 +426,9 @@ class MultiLegalPile(datasets.GeneratorBasedBuilder):
                         for name, train_val_files in _PILE_OF_LAW_NAMES[type].items():
                             if name == 'canadian_decisions':
                                 info["jurisdiction"] = "Canada"
-                            elif name in ['echr']:
+                            elif name in ['echr', 'eurlex', 'euro_parl']:
                                 info["jurisdiction"] = "EU"
-                            elif name in ['constitutions']:
+                            elif name in ['constitutions', 'icj-pcij', 'r_legaladvice', 'un_debates', 'tos']:
                                 info["jurisdiction"] = "N/A"
                             for split in ["train", "validation"]:
                                 for file_name in train_val_files[split]:

README.md

@@ -1,44 +1,45 @@
 ---
 annotations_creators:
+
 - other
-language_creators:
+  language_creators:
 - found
-language:
-- bg 
-- cs 
-- da 
-- de 
-- el 
+  language:
+- bg
+- cs
+- da
+- de
+- el
 - en
-- es 
-- et 
-- fi 
-- fr 
+- es
+- et
+- fi
+- fr
 - ga
-- hr 
-- hu 
-- it 
-- lt 
-- lv 
+- hr
+- hu
+- it
+- lt
+- lv
 - mt
-- nl 
-- pl 
-- pt 
-- ro 
-- sk 
-- sl 
+- nl
+- pl
+- pt
+- ro
+- sk
+- sl
 - sv
-license:
+  license:
 - cc-by-4.0
-multilinguality:
+  multilinguality:
 - multilingual
-paperswithcode_id: null
-pretty_name: "MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain"
-size_categories:
+  paperswithcode_id: null
+  pretty_name: "MultiLegalPile: A Large-Scale Multilingual Corpus for the Legal Domain"
+  size_categories:
 - 10M<n<100M
-source_datasets:
+  source_datasets:
 - original
-task_categories:
+  task_categories:
 - fill-mask
 
 ---
@@ -74,15 +75,15 @@ task_categories:
 ## Dataset Description
 
 - **Homepage:**
-- **Repository:** 
-- **Paper:** 
+- **Repository:**
+- **Paper:**
 - **Leaderboard:**
 - **Point of Contact:** [Joel Niklaus](mailto:joel.niklaus.2@bfh.ch)
 
 ### Dataset Summary
 
 The Multi_Legal_Pile is a large-scale multilingual legal dataset suited for pretraining language models.
-It spans over 24 languages and four legal text types. 
+It spans over 24 languages and five legal text types.
 
 ### Supported Tasks and Leaderboards
 
@@ -96,21 +97,25 @@ ro, sk, sl, sv
 ## Dataset Structure
 
 It is structured in the following format:
-text_type -> language -> jurisdiction.jsonl.xz
+type -> language -> jurisdiction.jsonl.xz
 
-text_type is one of the following:
+type is one of the following:
 
 - caselaw
 - contracts
 - legislation
 - other
+- mc4_legal
 
+`mc4_legal` is a subset of the other type but is listed separately so it can be easily excluded since it is less
+permissively licensed than the other types.
 
 Use the dataset like this:
+
 ```python
 from datasets import load_dataset
 
-config = 'en_contracts' # {language}_{text_type}
+config = 'en_contracts'  # {language}_{type}
 dataset = load_dataset('joelito/Multi_Legal_Pile', config, split='train', streaming=True)
 ```
 
@@ -122,17 +127,16 @@ all_legislation').
 
 The file format is jsonl.xz and there is one split available ("train").
 
+The complete dataset (653GB) consists of four large subsets:
 
-The complete dataset (564GB) consists of four large subsets:
-- Native Multi Legal Pile (30GB)
+- Native Multi Legal Pile (49GB)
 - Eurlex Resources (179GB)
 - MC4 Legal (133GB)
-- Pile of Law (222GB)
-
+- Pile of Law (292GB)
 
 #### Native Multilingual Legal Pile data
 
-| Language | Text Type   | jurisdiction | Source                             | Size (MB) | Tokens | Documents | Words/Document | URL | License |
+| Language | Text Type   | Jurisdiction | Source                             | Size (MB) | Tokens | Documents | Words/Document | URL | License |
 |:---------|:------------|:-------------|:-----------------------------------|----------:|-------:|----------:|---------------:|-----|--------:|
 | bg       | legislation | Bulgaria     | MARCELL                            |       588 |    xxx |       xxx |            xxx |     |         |
 | cs       | caselaw     | Czechia      | CzCDC Constitutional Court         |       713 |    xxx |       xxx |            xxx |     |         |
@@ -158,10 +162,11 @@ The complete dataset (564GB) consists of four large subsets:
 | nl       | legislation | Belgium      | ejustice                           |       178 |    xxx |       xxx |            xxx |     |         |
 | pl       | legislation | Poland       | MARCELL                            |       264 |    xxx |       xxx |            xxx |     |         |
 | pt       | caselaw     | Brazil       | RulingBR                           |       173 |    xxx |       xxx |            xxx |     |         |
+| pt       | caselaw     | Brazil       | CRETA                              |     19000 |    xxx |       xxx |            xxx |     |         |
 | ro       | legislation | Romania      | MARCELL                            |      2704 |    xxx |       xxx |            xxx |     |         |
 | sk       | legislation | Slovakia     | MARCELL                            |       192 |    xxx |       xxx |            xxx |     |         |
 | sl       | legislation | Slovenia     | MARCELL                            |       753 |    xxx |       xxx |            xxx |     |         |
-| total    | all         | all          | all                                |     29759 |    xxx |       xxx |            xxx |     |         |
+| total    | all         | all          | all                                |     48759 |    xxx |       xxx |            xxx |     |         |
 
 #### Eurlex Resources
 
@@ -175,15 +180,6 @@ See [MC4 Legal](https://huggingface.co/datasets/joelito/mc4_legal#data-instances
 
 See [Pile-of-Law](https://huggingface.co/datasets/pile-of-law/pile-of-law#data-instances) for more information.
 
-For simplicity and with respect to balancing off data across different jurisdictions and languages,
-we disregard many US resources that are either very specialized (e.g., tax rulings),
-outdated/historical (e.g., founding letters),
-very small (less than 20MB),
-not legal language in the strict sense (conversations),
-or the information overlaps with other sources (e.g. study materials or eurlex).
-If you are interested in a US-based (US-biased) model, refer to the "Pile of (US) Law" by Henderson et al. (2022).
-Analyses are put into the "other" category because in mc4_legal we also likely have similar text.
-
 | Language   | Type        | Jurisdiction   | Source                               |   Size (MB) |      Tokens |   Documents |   Tokens/Document | Part of Multi_Legal_Pile   |
 |:-----------|:------------|:---------------|:-------------------------------------|------------:|------------:|------------:|------------------:|:---------------------------|
 | 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
 
 [More Information Needed]
 
-
 ### Annotations
 
 #### Annotation process