RevolutionCrossroads/nara_revolutionary_war_pension_files_PDFs

Dataset Card for American Revolutionary War Pension Files - File-Level

---

Dataset Summary

A dataset derived from the National Archives and Records Administration (NARA) series Case Files of Pension and Bounty-Land Warrant Applications Based on American Revolutionary War Service (NARA Catalog Series, NAID 300022). This dataset provides a file-level representation of Revolutionary War pension records, aggregating individual page records into complete pension files with associated metadata, extracted text, transcriptions, and and AI-generated OCR outputs.

It is derived from the page-level dataset:
https://huggingface.co/datasets/RevolutionCrossroads/nara_revolutionary_war_pension_files

This dataset was prepared as part of the Revolution Crossroads project:
https://www.si.edu/revolution-crossroads

---

Dataset Description

For the initiative Revolution Crossroads, the Smithsonian Institution prepared this dataset using data, metadata, and digital objects publicly available from the National Archives Catalog.

Dataset Details

• Prepared by: Smithsonian Institution, Office of Digital & Innovation staff
• Shared by: Revolution Crossroads
• Language(s): English, some French
• License: Public domain

---

Relationship to Source Dataset

This dataset is a transformation of the original Revolution Crossroads page-level dataset. For detailed information about:

• source materials
• digitization processes
• metadata definitions
• rights and provenance

refer to the page-level dataset card:
https://huggingface.co/datasets/RevolutionCrossroads/nara_revolutionary_war_pension_files/blob/main/README.md

---

Curation Rationale

The original dataset is structured at the page level, where each record represents a single page from a pension file. However, pension files are multi-page archival documents that often contain narratives, affidavits, correspondence, and supporting materials that span multiple pages.

This file-level dataset was created to:

• Preserve document-level context across entire pension files
• Support OCR and text extraction workflows at the file level
• Enable analysis of relationships between individuals, places, and events within a single case file
• Support entity extraction and historical interpretation across full narrative records

---

Dataset Creation

This dataset was derived from the page-level pension dataset by restructuring records from page-level to file-level.

• Page-level records were grouped by NAID (file-level identifier)
• File-level records retain key metadata from the source dataset
• Page counts were calculated for each file (numberOfPages)
• Page-level data (images, extracted text, and transcriptions) were retained as lists within each file-level record
• Some page-level metadata fields not relevant to file-level analysis were removed

A note on archival levels:

• In most cases, records correspond to file-level groupings. However, a small number of records represent an intermediate “item” level between file and page in the NARA Catalog hierarchy
• These item-level records were treated equivalently to file-level records in this dataset
• Any record containing one or more pages was used as the top-level grouping unit (NAID), regardless of whether it was formally classified as a file or item

Each record in this dataset corresponds to a single pension file or equivalent document-level unit.

---

Data Collection and Processing

Supporting Files Available

File-level PDF files are provided in the Files tab for this dataset, within the media directory.

• pdfs
  Contains compiled PDF files representing complete pension files. Each PDF corresponds to a single record in the dataset and is constructed from the source page images.

Note:
pdfObjectID and pdfURL fields refer to PDFs provided by the National Archives Catalog where available. These may be incomplete or inconsistently available. The supplementary PDFs included with this dataset were generated to provide consistent, complete file-level representations. Some PDFs for newly added files are forthcoming.

PDF Construction

• Source page images were obtained as JPG files from the NARA Catalog
• Images were grouped by top-level NAID and compiled into multi-page PDFs
• PDFs were constructed to represent complete pension files

To ensure compatibility with downstream processing systems:

• Some PDFs were resized when they exceeded size limits during OCR extraction workflows
• Resizing was applied selectively to affected files
• A future update will provide consistently resized PDFs for broader external use

AI OCR Enrichment

This dataset has been enriched with outputs generated by DataLab's Chandra OCR 2, a Vision Language Model (VLM) for document understanding.

Complete Chandra OCR 2 processing outputs are stored in the Revolution Crossroads VLM OCR Hugging Face bucket. The bucket includes a README describing the processing pipeline, output formats, and directory structure.

• Complete VLM OCR Bucket: RevolutionCrossroads_VLM_OCR
• Revolutionary War Pension Files: rev_war_pension

Two additional fields have been added:

• RevX_OCR_bucket provides a link to the complete Chandra OCR 2 processing outputs for a document. These outputs include structured JSON, Markdown, HTML, layout information, extracted images, quality metrics, and other artifacts generated during processing.

• RevX_OCR_text contains the page-level plain text extracted by Chandra OCR 2. The text is provided inline to simplify downstream analysis while the complete processing outputs remain available through the linked bucket.

These outputs represent the direct results of automated AI processing. They have not been manually validated or corrected and should be treated as machine-generated data.

Not every record in this dataset currently includes Chandra OCR 2 outputs. The initial processing, completed in March 2026, focused on Revolutionary War Pension Files that had not yet been human transcribed at the time of processing. Since then, the source collections have been refreshed with additional materials and new transcriptions. A second round of Chandra OCR 2 processing is currently underway, and additional processing outputs will be incorporated into future dataset releases.

Accessing a Specific Data Version

This dataset may be updated over time as source data is revised, normalized, or restructured. If you are using the data for a project, we recommend accessing a specific tagged version so your workflow remains stable and reproducible.

You can access a tagged version using the revision parameter in load_dataset:

from datasets import load_dataset
import polars as pl

dataset_name = "RevolutionCrossroads/nara_revolutionary_war_pension_files_PDFs"
commit_tag = "Source_2026-04-10"

ds = load_dataset(
    dataset_name,
    revision=commit_tag,
    split="train"
)

# Convert to pandas
pandas_df = ds.data.table.to_pandas()

# Convert to Polars
polars_df = pl.from_arrow(ds.data.table)

This example loads the specified tagged dataset version and converts it to either a pandas or Polars dataframe for analysis.

Available Data Versions

• Source_2026-04-10 – Initial PDF-level dataset with refreshed source data.
• new_OCR_v1 – Adds Datalab AI-generated OCR while preserving the PDF-level dataset structure.

Using a tagged version is recommended for research, publication, or project workflows where consistency over time is important.

---

Dataset Structure

Each record in the dataset corresponds to a single pension file. File-level metadata is represented once per record, while page-level data (images, text, transcriptions) are stored as lists.

---

Data Fields

• NAID (string)
  National Archives Identifier for the pension file.
  Example:
  111403815

• naraURL (string)
  Link to the pension file in the National Archives Catalog.
  Example:
  https://catalog.archives.gov/id/111403815

• title (string)
  Title of the pension file.
  Example:
  Revolutionary War Pension and Bounty Land Warrant Application File B. L. Wt. 2,153-400, Joseph Torrey

• logicalDate (string)
  Normalized or machine-readable date, if available.
  Example:
  1836-06-02

• pageObjectId (list)
  List of identifiers for individual pages within the file.
  Example:
  ["111403816", "111403817"]

• pageURL (list)
  List of URLs to page images.
  Example:
  ["https://s3.amazonaws.com/NARAprodstorage/.../image1.jpg"]

• numberOfPages (int64)
  Number of pages in the pension file.
  Example:
  5

• pageImageType (string)
  File format of the page images.
  Example:
  JPG

• fs_extractedTextID (list)
  Identifiers for Family Search extracted text records.
  Example:
  ["19537435", "19537436"]

• fs_extractedText (list)
  Extracted text for each page, provided via automated OCR/AI processes by Family Search.
  Example:
  ["State of New York...", "City and County of New York..."]

• fs_extractedTextDate (list)
  Dates when Family Search extracted text was created.
  Example:
  ["2024-11-20T13:30:07.000Z"]

• transcriptionID (list)
  Identifiers for human transcription records, if available.
  Example:
  ["60f6e82a-f53d-4de3-a09b-f4f707046c0f"]

• transcriptionText (list)
  Human-created transcription text for pages, where available.
  Example:
  ["Gorham Me March 22 1868..."]

• transcriptionIngestDate (list)
  Dates when human transcriptions were added to the catalog.
  Example:
  ["2025-04-30 17:15:44"]

• pdfURL (string)
  Redirect URL to the file-level PDF as stored on Hugging Face in this dataset. Note: Some PDFs for newly added files are forthcoming, but links have been predetermined. Example:
  https://huggingface.co/datasets/RevolutionCrossroads/nara_revolutionary_war_pension_files_PDFs/resolve/main/media/pdfs/54961568-111412574/111403760.pdf

• RevX_OCR_bucket (string)
  Link to the complete Chandra OCR 2 processing outputs for this document. These outputs include structured representations, layout information, extracted images, and additional processing artifacts. May be null if the document has not yet been processed. Example:
  https://huggingface.co/buckets/RevolutionCrossroads/RevolutionCrossroads_VLM_OCR/tree/rev_war_pension/...

• RevX_OCR_text (list)
  Page-level text extracted by DataLab's Chandra OCR 2. One list entry corresponds to each page of the document. May be null if the document has not yet been processed. Example:
  ["SERVICE\nNUMBER...", "State of New York..."]

---

Source Data

This dataset is derived from the National Archives and Records Administration series:
https://catalog.archives.gov/id/300022

For full details on the source collection, refer to the page-level dataset card:
https://huggingface.co/datasets/RevolutionCrossroads/nara_revolutionary_war_pension_files/blob/main/README.md

---

Personal and Sensitive Information

No known personal or sensitive information is included beyond what appears in historical archival records. These materials may contain outdated or offensive terminology reflective of the period in which they were created.

---

Considerations for Using the Data

Risks and Limitations

• Pension files vary widely in length, structure, and completeness
• Extracted text accuracy varies, especially for handwritten documents
• Transcriptions are available for only a portion of records
• Metadata may be incomplete or inconsistent

Recommendations

Researchers should validate extracted information against source images when accuracy is critical and consult the National Archives Catalog for the most up-to-date records.

---

Additional Information

Citation Information

BibTeX

@misc{revolution_crossroads_2026,
    author       = { Revolution Crossroads },
    title        = { nara_revolutionary_war_pension_files_PDFs (Revision 8453ff3) },
    year         = 2026,
    url          = { https://huggingface.co/datasets/RevolutionCrossroads/nara_revolutionary_war_pension_files_PDFs },
    doi          = { 10.57967/hf/8349 },
    publisher    = { Hugging Face }
}

APA

Revolution Crossroads Project Team. (2025). American Revolutionary War Pension Files - File-Level [Data set]. Hugging Face.

---

Glossary

• NAID: National Archives Identifier
• Extracted text: Machine-generated text created from images
• Transcription: Human-created text
• Pension file: A complete archival case file documenting a pension application

---

Dataset Card Contact

revolutioncrossroads@si.edu