---
language: en
license: apache-2.0
viewer: false
---

# Dataset card for Historical English Books

This dataset is a collection of historical English books from before `1919`. It aggregates books from the following datasets into a single dataset.

- [British Library Books](https://huggingface.co/datasets/TheBritishLibrary/blbooks)
- [Biodiversity Heritage Library](https://huggingface.co/datasets/common-pile/biodiversity_heritage_library_filtered)
- [Library of Congress "Selected Digitized Books"](https://huggingface.co/datasets/common-pile/library_of_congress_filtered)
- [Project Gutenberg PG19](https://huggingface.co/datasets/deepmind/pg19)

A blend of each of the datasets was selected using the following rules.

- Language = "English"
- Limit natural science to most popular publications (>1 edition) from Biodiversity Heritage Library
- Select only science subjects from Library of Congress collection given good general literature coverage from other sources

## Fields

The following fields are available.

| Field       | Description                                        |
| ----------- | -------------------------------------------------- |
| id          | Book id                                            |
| source      | Source collection (blbooks, bhl, loc, pg19)        |
| year        | Publication year                                   |
| title       | Book title                                         |
| abstract    | LLM-generated abstract written in prose of 1800s   |
| label       | [Domain category](https://huggingface.co/NeuML/domain-labeler) applied to book                    |
| quality     | OCR quality (0.0 - 1.0)                            |
| text        | Full text of the book                              |

## Disclaimer

Note that these books represent cultural values at the time of publishing which are often not in line with our modern values.