dataset_info:
features:
- name: opinion_id
dtype: int64
- name: opinion
dtype: string
- name: opinion_source
dtype: string
- name: opinion_word_count
dtype: int64
- name: opinion_date_created
dtype: string
- name: opinion_type
dtype: string
- name: opinion_extracted_by_ocr
dtype: string
- name: opinion_per_curiam
dtype: string
- name: cluster_id
dtype: int64
- name: cluster_judges
dtype: string
- name: cluster_nature_of_suit
dtype: string
- name: cluster_source
dtype: string
- name: cluster_blocked
dtype: string
- name: cluster_precedential_status
dtype: string
- name: cluster_citation_count
dtype: int64
- name: cluster_case_name
dtype: string
- name: cluster_case_name_short
dtype: string
- name: cluster_case_name_full
dtype: string
- name: cluster_summary
dtype: string
- name: cluster_history
dtype: string
- name: cluster_headmatter
dtype: string
- name: cluster_headnotes
dtype: string
- name: cluster_posture
dtype: string
- name: cluster_arguments
dtype: string
- name: cluster_cross_reference
dtype: string
- name: cluster_disposition
dtype: string
- name: cluster_syllabus
dtype: string
- name: docket_id
dtype: int64
- name: docket_number
dtype: string
- name: docket_view_count
dtype: int64
- name: court_id
dtype: string
- name: court_jurisdiction
dtype: string
- name: court_in_use
dtype: string
- name: court_short_name
dtype: string
- name: court_full_name
dtype: string
splits:
- name: train
num_bytes: 5849416
num_examples: 503
- name: test
num_bytes: 4916437
num_examples: 450
download_size: 5617493
dataset_size: 10765853
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
language:
- en
tags:
- legal
Dataset Card for Free-Law-Project/opinions-metadata
This dataset contains the metadata of case law opinions used for training and evaluating the Free Law Project Semantic Search Project.
Dataset Details
Dataset Description
- Curated by: Free Law Project
- Language(s) (NLP): English
The dataset is curated by Free Law Project by randomly sampling ~1K cases across various courts and jurisdictions from the CourtListener database.
This dataset contains all the metadata associated with each opinion, with opinion_id as the unique identifier. The train split is used for finetuning the embedding semantic search models while the test split is used for evaluating the performance of various embedding semantic search models.
Uses
This rich dataset can be used for pretraining and finetuning various legal tasks, such as classification, semantic search, summarization, etc. Due to wide range of opinion length, this dataset can be particularly valuable for long context use cases.
Refer to the CourtLisener Case Law API Doc for detailed information on the features.
Our data is vast and this is only a fraction of the Case Law data we have.
Citation
If you use this data, please cite with the following entry:
@misc{Free Law Project,
title={Free law project: Making the legal ecosystem more equitable and competitive.},
url={https://free.law/},
author={Free Law Project}}