hyperdemocracy/usc-textversion

Dataset Description

This dataset is part of a family of datasets that provide convenient access to congressional data from the US Government Publishing Office via the GovInfo Bulk Data Repository. GovInfo provides bulk data in xml format. The raw xml files were downloaded using the congress repo. Further processing was done using the legisplain legisplain repo.

Hyperdemocracy Datasets

• usc-billstatus (metadata on each bill)
• usc-textversion (different text versions of bills in xml)
• usc-unified (combined metadata and text version xml)

TEXTVERSIONS (text for congresses 113-119)

• https://www.govinfo.gov/bulkdata/BILLS
• https://xml.house.gov/
• https://github.com/usgpo/bill-dtd?tab=readme-ov-file

These xml files contain multiple text versions for each bill.

Column Descriptions

Column	Description
tv_id	a unique ID for each text version ({congress_num}-{legis_type}-{legis_num}-{legis_version}-{xml_type})
legis_id	a unique ID for each bill ({congress_num}-{legis_type}-{legis_num})
congress_num	the congress number for the bill
legis_type	one of [hr, hres, hconres, hjres, s, sres, sconres, sjres] (see govinfo - types of legislation)
legis_num	bills in each congress and of each type get an incrementing number as part of their ID
legis_version	version of bill text (see govinfo - common versions of bills)
legis_class	one of [bills, plaw]
bulk_path	XML file path during bulk download
file_name	last part of bulk_path. used in joining to billstatus
lastmod	lastmod date during bulk download
xml_type	one of [dtd, uslm]
root_tag	the root xml tag. one of [bill, resolution, amendment-doc, pLaw]
tv_xml	contents of textversion XML file
tv_txt	a plain text version of the XML content

Examples

The dataset is broken into splits (one split per congress number).

from datasets import load_dataset

# load each split into a `DatasetDict` keyed on congress number
dsd = load_dataset(path="hyperdemocracy/usc-textversion")

# load a single congress number into a `Dataset`
ds = load_dataset(path="hyperdemocracy/usc-textversion", split=117)

# load all congress numbers into a single `Dataset`
ds = load_dataset(path="hyperdemocracy/usc-textversion", split="all")

Congress Number to Date Mapping

Congress Number	Years	Metadata	Text
119	2025-2026	True	True
118	2023-2024	True	True
117	2021-2022	True	True
116	2019-2020	True	True
115	2017-2018	True	True
114	2015-2016	True	True
113	2013-2014	True	True