HTRC data access

Learn about the different access points and formats to the data HTRC provides, as well as the various affordances and limitations of each method. Your research project will largely dictate which method is best suited for your needs.  

HTRC provides access to data from the HathiTrust corpus in several forms across its suite of tools and services for computational text analysis. Data is periodically synced from HathiTrust, but not all HTRC tools and services are updated on the same schedule. Additionally, copyright, user agreements, and security concerns impact data availability and format. 

  • HTRC algorithms and HTRC Data Capsules are capable of analyzing the entire HathiTrust corpus, and additionally make use of each volume’s MARC bibliographic and METS metadata. Both the HTRC algorithms and Capsule-environments draw from the HTRC Data API described below.

  • The HTRC makes available also two datasets, the HTRC Extracted Features Dataset and a dataset of Word Frequencies in English Language Literature, 1700-1922. HTRC Extracted Features includes metadata and extracted page-level data (words and word counts) for 15.7 million volumes.

  • HathiTrust+Bookworm visualizes data for 15.7 million volumes.

HathiTrust text data

Access options

The following table disambiguates textual data access, including availability and format, within the HTRC ecosystem. 


Tool or service

# volumes available (as of 8/19)

rights statusdata access mechanismfile formatdata formatpermissions required
HTRC Analytics algorithms17 millionAllVia HTRC Workset; researcher runs tool without accessing underlying data(no file access)

Uncorrected OCR text data; only results are exposed to researcher (not the underlying data)

HTRC Analytics account
HT+Bookworm tool15.7 millionAllVia web-interface; researcher visualizes data without accessing underlying data(no file access)Unigrams (single words), based on HTRC Extracted Features dataset; underlying data not exposed to researcher(none)
HTRC Data Capsule6.5 million for everyone; 17 million for affiliates of HT member institutions upon approval or requestPublic domain for everyone; In-copyright limited to affiliates of HT member institutions upon approval of request/wiki/spaces/COM/pages/43286551Zipped text files in PairTree directory structureUncorrected OCR text dataHTRC Analytics account
HTRC Extracted Features dataset15.7 millionAllrsyncJSON files in PairTree directory structureVolume- and page-level metadata and part-of-speech tagged page-level "bags of words"(none)
HathiTrust dataset request6.5 million, dependent on institutional agreementsPublic domain; accessibility of Google-digitized volumes dependent on whether researcher's home institution has signed agreementrsync

Zipped text files in PairTree directory structure

Uncorrected OCR text dataCustom dataset request application

Building a workset/dataset of HathiTrust text data

HTRC Data API

This table describes the HTRC Data API.



 

HTRC Data API

purposeto provide access to HathiTrust text data within an HTRC Data Capsule AND to serve high-performance large-scale algorithms and programs (not publicly-accessible)
data availableentire HathiTrust corpus
useIn-capsule via the HTRC Workset Toolkit
throttling enforcementno
securityJWT
bulk retrieval of volumesyes
metadata availableMETS