HathiTrust Research Center receives NEH support for open research tools

Stephen Downie
J. Stephen Downie, Professor, Associate Dean for Research, and Co-Director of the HathiTrust Research Center

The HathiTrust Research Center (HTRC), cohosted by the iSchool at Illinois and the Luddy School of Informatics at Indiana University, has received a $325,000 Digital Humanities Advancement Grant from the National Endowment for the Humanities. One of 15 awarded nationwide, this grant will support the development of a new set of visualizations, analytical tools, and infrastructure to enable users to interact more directly with the rich data extracted from the HathiTrust Digital Library’s collection of more than 17.5 million digitized volumes.

The project, "Tools for Open Research and Computation with HathiTrust: Leveraging Intelligent Text Extraction" (TORCHLITE) will be led jointly by Professor J. Stephen Downie, associate dean for research at the iSchool and co-director of HTRC, and John Walsh, HTRC director and associate professor of information and library science at Indiana University. HTRC staff at both universities will collaborate during the next two years to accomplish TORCHLITE's goals.

"TORCHLITE will enable us to increase dramatically, and to open more fully, public access to the massive, rich data that HTRC has created from the HathiTrust Digital Library corpus," said Downie. "We have already developed innovative ways to transform, enhance, and provide access to data created from the many millions of scanned books held by HathiTrust. With TORCHLITE, we'll create new methods for accessing this data, together with several easy-to-use tools to allow people to interact with it, analyze it, and visualize it in novel ways."

The data of interest is contained in HTRC's flagship "Extracted Features" (EF) dataset, which consists of rich metadata and statistical information inferred by algorithm from the digitized texts of the entire HathiTrust corpus and documents every word on every page, including the number of times the word appears, its part of speech, and other formal features of the language on the page. The EF dataset, and methods for computing over it, have enabled many forms of full-text analysis—even of copyrighted materials. The EF dataset contains nearly three trillion tokens (or in other words, words) representing more than six billion pages of text, making it arguably the largest open dataset of its kind that is readily available to researchers around the world.

In addition to creating new methods for dealing with this enormous data set, and perhaps more impactfully, Downie emphasized that TORCHLITE will develop a framework on which digital humanities scholars, digital librarians, data scientists, and anyone else interested in textual analysis, can build and implement their own tools, and deploy them in their own environments, in order to access the HTRC data more directly and openly. TORCHLITE's tools and methods will enable the retrieval of standard volume-level descriptors—such as title, publisher, date of publication, genre, and page count—along with page- and word-level linguistic and statistical information.

In addition to creating interactive, easy-to-use tools and dashboards, TORCHLITE will promote broad community engagement through a workshop and a mentored hackathon in autumn 2023, in the hopes of encouraging individual researchers to develop their own tools using the project’s application programming interface (API).

HTRC is the official research arm of HathiTrust, a library consortium that hosts books owned and digitized by its member libraries, often in cooperation with the Google Books Project and other mass-digitization efforts. Its mission is to contribute to the common good by collecting, organizing, preserving, communicating, and sharing the record of human knowledge through data-intensive computational methods.

Updated on
Backto the news archive

Related News

Winning exhibit features recipes from across the globe

MSLIS students Yung-hui Chou, Alice Tierney-Fife, and Elizabeth Workman are the winners of this year’s Graduate Student Exhibit Contest, sponsored by the University of Illinois Library. Their exhibit, "Culture and Cuisine in Diaspora: A Hidden Library Collection," displays items from seven campus libraries and highlights research and recreational material centered on traditional recipes from across the globe. The exhibit is on display in the library's Marshall Gallery through the end of April and also available online.

Trainor receives the Karen Wold Level the Learning Field Award

Senior Lecturer Kevin Trainor has been selected by the Division of Disability Resources and Educational Services (DRES) to receive the 2024 Karen Wold Level the Learning Field Award. This award honors exemplary members of faculty and staff for advocating and/or implementing instructional strategies, technologies, and disability-related accommodations that afford students with disabilities equal access to academic resources and curricula. 

Kevin Trainor

Seo coauthors chapter on data science and accessibility

Assistant Professor JooYoung Seo and Mine Dogucu, professor of statistics in the Donald Bren School of Information and Computer Sciences at the University of California Irvine, have coauthored a chapter in the new book Teaching Accessible Computing. The goal of the book, which is edited by Alannah Oleson, Amy J. Ko and Richard Ladner, is to help educators feel confident in introducing topics related to disability and accessible computing and integrating accessibility into their courses.

JooYoung Seo

iSchool instructors ranked as excellent

Fifty-five iSchool instructors were named in the University's List of Teachers Ranked as Excellent for Fall 2023. The rankings are released every semester, and results are based on the Instructor and Course Evaluation System (ICES) questionnaire forms maintained by Measurement and Evaluation in the Center for Innovation in Teaching and Learning. 

iSchool Building

New study shows LLMs respond differently based on user’s motivation

A new study conducted by PhD student Michelle Bak and Assistant Professor Jessie Chin, which was recently published in the Journal of the American Medical Informatics Association (JAMIA), reveals how large language models (LLMs) respond to different motivational states. In their evaluation of three LLM-based generative conversational agents (GAs)—ChatGPT, Google Bard, and Llama 2—the researchers found that while GAs are able to identify users' motivation states and provide relevant information when individuals have established goals, they are less likely to provide guidance when the users are hesitant or ambivalent about changing their behavior.