Dataset Card for LegalOps JustiaCorpus

Dataset Summary

This is an English-Language dataset consisting of US Supreme and Federal Court cases, with fulltexts and summaries. The dataset comprises approximately 600 Supreme court cases with Summaries and Syllabi, as well as those with missing data scraped from https://supreme.justia.com/.

As the highest court in the nation, the U.S. Supreme Court has shaped the rights and freedoms of Americans since the Founding. Justia provides a free collection of all U.S. Supreme Court decisions from 1791 to the present.

The federal court data is scraped from https://law.justia.com/cases/federal/, it is more sparse than the Supreme Court data, with approximately 17,000 Cases out of a total of 284,000 scraped from the records with non-zero length Fulltext and Summaries.

Dataset Split	Number of Rows
federal	284011
federal-clean	16818
supreme	1022
supreme-clean	593

Supported Tasks and Leaderboards

• Text Summarisation

Languages

• English - (n.b. - The BCP-47 code for English as generally spoken in the United States is en-US)

Dataset Structure

Data Instances

The data instances for the splits differ slightly between the Federal and Supreme Court data

Supreme Court Data

Each instance in the "supreme" consists of a string fulltext field containing the Syllabus of the case (Syllabus), a string summary field containing a human-written summary of the case. There are two other significant data fields for the case, one containing the opinion (Opinion), the other (optionally) the Dissent (Dissent). The other fields are metadata for the case scrape, a string tag for identifiying "federal" or "supreme" court should the splits be combined (tag), the scraped URL (url), a list of a dictionary of metadata for the case itself (metadata), a list of file URLs (file_urls) and a more complete data structure (files) consisting of a list of dictionaries for each PDF downloaded. The "path" key in this datastructure relates to the location within the associate PDF tarball (see (# Additional Data)[# Additional Data])

{
  "Syllabus": "See United States v. Detroit Timber & Lumber Co.,\n\n200 U.S. 321, 337.\nSUPREME COURT OF THE UNITED STATES\nSyllabus\nGREENE, aka TRICE v. FISHER, SUPERINTENDENT, STATE CORRECTIONAL INSTITUTION AT SMITHFIELD, et al.\ncertiorari to the united states court of appeals for the third circuit\nNo. 10\u2013637.\u2003Argued October 11, 2011\u2014Decided November 8, 2011\nDuring petitioner Greene\u2019s trial for murder, robbery, and conspiracy, the prosecution introduced the redacted confessions of two of Greene\u2019s nontestifying codefendants.  A jury convicted Greene.  The Pennsylvania Superior Court...",
  "Dissent": "",
  "Opinion": "SUPREME COURT OF THE UNITED STATES\n_________________\nNo. 10\u2013637\n_________________\nERIC GREENE, aka JARMAINE Q. TRICE, PETI- TIONER v. JON FISHER, SUPERINTENDENT, STATE CORRECTIONAL INSTITUTION AT SMITHFIELD, et al.\non writ of certiorari to the united states court of appeals for the third circuit\n[November 8, 2011]\n\nJustice Scalia delivered the opinion of the Court.\nUnder the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), a federal court may not grant habeas relief to a state prisoner...,
  "tag": "supreme",
  "url": "https://supreme.justia.com/cases/federal/us/565/34/",
  "file_urls": [
    "https://supreme.justia.com/cases/federal/us/565/10-637/case.pdf"
  ],
  "files": [
    {
      "checksum": "7364db9dec242c4bf751cddd1082c714",
      "path": "full/aff7f0b60e06bcdc14db0962db7a187460cf3d6e.pdf",
      "status": "downloaded",
      "url": "https://supreme.justia.com/cases/federal/us/565/10-637/case.pdf"
    }
  ],
  "metadata": {
    "Advocates": null,
    "Argued": "October 11, 2011",
    "Decided": "November 8, 2011",
    "Docket No.": "10-637",
    "First Party": "Eric Greene, aka Jarmaine Q. Trice",
    "Granted": "April 4, 2011",
    "Juris Postponed": null,
    "Official Citation": "565 U.S. 34",
    "Reargued": null,
    "Second Party": "Jon Fisher, Superintendent, State Correctional Institution at Smithfield, et al.",
    "page": 34,
    "volume": 565
  },
}

Federal Court Data

Each instance consists of a string fulltext field containing the fulltext of the case, a string summary field containing a human-written summary of the case. The other fields are metadata for the case scrape, a string tag for identifiying "federal" or "supreme" court should the splits be combined (tag), the scraped URL (url), a list of a dictionary of metadata for the case itself (metadata), a list of file URLs (file_urls) and a more complete data structure (files) consisting of a list of dictionaries for each PDF downloaded. The "path" key in this datastructure relates to the location within the associate PDF tarball (see (# Additional Data)[# Additional Data])

{
  "fulltext": "Appeal from judgment of the United States District Court for the Western District of New York (Telesca, J.). The district court denied Petitioner habeas corpus relief after finding that Petitioner did not derive citizenship from his father; the district court ruled that Petitioner was not in his father    s     legal custody     when his father naturalized. We conclude that the district court erred because it relied on an unenforceable custody award.     Legal custody ...",
  "summary": "Petitioner appealed from the district court's denial of habeas corpus relief after finding that he did not derive citizenship from his father. The district court ruled that petitioner was not in his father's \"legal custody\" when his father naturalized. The court concluded that the district court erred because it relied on an unenforceable Dominican Republic custody award where New York had jurisdiction to determine custody. Accordingly, the court vacated the judgment and remanded for further proceedings.",
  "tag": "federal",
  "url": "https://law.justia.com/cases/federal/appellate-courts/ca2/09-4211/09-4211_opn-2011-12-29.html",
  "file_urls": [
    "https://cases.justia.com/federal/appellate-courts/ca2/09-4211/09-4211_opn-2011-12-29.pdf"
  ],
  "files": [
    {
      "checksum": "c48f9dd5a186a0e4dde4259085d99840",
      "path": "full/e38b7ce3ff0e4f83a100f5e2cc57552591d033b0.pdf",
      "status": "downloaded",
      "url": "https://cases.justia.com/federal/appellate-courts/ca2/09-4211/09-4211_opn-2011-12-29.pdf"
    }
  ],
  "metadata": {
    "court_id": "ca2",
    "date": "2011-12-29",
    "number": "09-4211",
    "title": "Garcia v. USICE (Dept. of Homeland Security), No. 09-4211 (2d Cir. 2011)"
  }
}

Data Fields

[More Information Needed]

Data Splits

This dataset is comprised of four splits: a source-quality version (federal/supreme)and a pre-processed version (federal-clean/supreme-clean). These are not currently further subdivided into train/test/eval splits. The default split is the supreme split.

JustiaCorpus/
├── README.md
├── federal/
├── federal-clean/
├── supreme/
└── supreme-clean/

The splits can be loaded as follows:

ds = load_dataset(
    "HartreeCentre/JustiaCorpus",
    split,
    token=HF_TOKEN_VALUE,
)["train"]

where split is one of ["supreme", "federal", "supreme-clean", "federal-clean"].

Additional Data

Within this repository, two tarballs containing the full PDF documents for each dataset can be found in the corresponding federal/ and supreme/ datasets.

• federal/federal_pdfs.tar.gz
• supreme/supreme_pdfs.tar.gz

to extract these, clone this dataset repo, navigate to the directory and untar:

tar -xzvf federal_pdfs.tar.gz

The mapping between case and PDF is stored in the "files" field in the dataset.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Personal and Sensitive Information

This dataset is not anonymized, so individuals' names can be found in the dataset

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

[More Information Needed]

Citation Information

@INPROCEEDINGS{9378308,
  author={Gargett, Andrew and Firth, Rob and Aletras, Nikolaos},
  booktitle={2020 IEEE International Conference on Big Data (Big Data)}, 
  title={LegalOps: A Summarization Corpus of Legal Opinions}, 
  year={2020},
  volume={},
  number={},
  pages={2117-2120},
  doi={10.1109/BigData50022.2020.9378308}}

Contributions

Thanks to @RobFirth for adding this dataset.