Skip to end of metadata
Go to start of metadata


Introduction

HTRC releases research datasets to facilitate text analysis using the HathiTrust Digital Library. While copyright-protected texts are not available for download from HathiTrust, fruitful research can still be performed on the basis of non-consumptive analysis of features extracted from full text. These features include volume-level metadata, page-level metadata, part-of-speech-tagged tokens, and token counts. Additionally, HTRC has partnered with advanced researchers to release a derived dataset, Word Frequencies in English-Language Literature, 1700-1922.

Getting Started

Downloading Extracted Features

Use Cases and Examples

Extracted Features in the Wild

Extracted Features Dataset 

Documentation

Get the data

Word Frequencies in English-Language Literature, 1700-1922 

Documentation

Get the data

Extracted Features Dataset [v.0.2]

NOTE: this dataset has been superseded by Extracted Features Dataset [v.1.0], above.

Documentation

Get the data

 

  • No labels