HathiTrust is often searched as a library catalog — a place to look up whether a book has been digitized. This guide is about a different, less-visible layer of HathiTrust: the HathiTrust Research Center (HTRC) and its non-consumptive research access model, which lets researchers computationally analyze the full HathiTrust corpus, including in-copyright volumes, without reading the underlying text sequentially the way a library patron would. If you came here looking for how to search or browse HathiTrust’s public catalog, see hathitrust.org directly; this page covers the computational-research and text-and-data-mining (TDM) access model specifically.
What HathiTrust Is
HathiTrust is a not-for-profit digital library partnership of academic and research libraries, founded in 2008 in response to large-scale library digitization efforts (including Google Books) as a way for member libraries to jointly preserve and provide long-term access to digitized print collections. It is administered from the University of Michigan on behalf of its member institutions. The public-facing HathiTrust Digital Library lets anyone search across millions of digitized volumes, read public-domain works in full, and view limited previews of in-copyright works — this is the “library catalog” use most searches for “HathiTrust” are actually looking for.
Separately from that public catalog, HathiTrust also functions as a large-scale digitized-text corpus: the underlying OCR (optical character recognition) text of every scanned volume, public-domain and in-copyright alike, exists as a dataset that can, under specific conditions, be analyzed computationally rather than read. That computational layer is what the HathiTrust Research Center provides.
What “Non-Consumptive Research” Means
HTRC defines non-consumptive research as computational analysis of text that does not substitute for a human reading the work in the way the work was intended to be read. In practice this means: extracting statistical or structural information (word frequencies, n-grams, part-of-speech patterns, topic models) from the corpus, rather than retrieving or displaying the original text in a form a person could sit down and read cover to cover. The distinction matters because it is the legal basis on which HTRC can offer this kind of access to in-copyright works at all — mass digitization and full-text search by HathiTrust were upheld as transformative fair use in Authors Guild v. HathiTrust (2d Cir. 2014), and the non-consumptive-research model is HTRC’s way of extending that same fair-use logic to computational text and data mining rather than just search.
The HathiTrust Research Center (HTRC)
HTRC is a research center jointly operated by Indiana University and the University of Illinois, in partnership with HathiTrust, that builds and hosts the computational infrastructure for non-consumptive research across the HathiTrust corpus. Its tools and datasets, made available through the HTRC Analytics platform, are the actual mechanism by which “text and data mining of HathiTrust” happens in practice. The main access paths are:
- HTRC Data Capsule — a secure, individually provisioned virtual research environment that gives a researcher non-consumptive computational access to the full OCR text of HathiTrust volumes, including in-copyright works. A capsule runs in a locked-down “secure mode” for work on in-copyright content: researchers can run algorithms against the full text inside the capsule, but cannot export the raw text itself — only aggregated, non-consumptive outputs (counts, models, derived features) are allowed to leave.
- HTRC Extracted Features Dataset — a freely downloadable dataset of page-level features (token and part-of-speech counts, language, structural metadata) for millions of HathiTrust volumes, covering both public-domain and in-copyright works. Because it never exposes running sequential text, it is non-consumptive by construction and does not require capsule access to use.
- Bookworm — a tool for visualizing word- and phrase-frequency trends across the corpus over time, conceptually similar to a large-scale n-gram viewer, built on the same non-consumptive derived data.
These are genuinely different tools for different research needs: the Extracted Features Dataset suits bulk, corpus-wide quantitative text analysis without needing to touch full text at all; the Data Capsule suits research that requires running custom algorithms directly against full in-copyright text, at the cost of a more restricted, vetted environment.
How This Differs From Browsing the HathiTrust Catalog
The distinction that matters for a researcher deciding where to start:
- HathiTrust Digital Library (the catalog) — public search and reading interface. Full-text reading access is limited to public-domain works (and a few other permitted categories); in-copyright works are searchable and show limited context, but are not readable in full. No special access request is needed to search or browse it.
- HTRC non-consumptive research tools — built for computational analysis at scale, not reading. They extend access to the underlying text of in-copyright works specifically because the output is not a substitute for reading them — it’s derived data or algorithm-mediated results. This access sits behind separate HTRC accounts, tooling, and (for Data Capsule secure-mode work) a review process, not the general HathiTrust login.
A researcher who wants to know “does HathiTrust have this book” is a catalog question. A researcher who wants to know “how does word usage in this genre change across a 200-year span of the corpus,” or “can I run topic modeling across a million in-copyright volumes,” is an HTRC non-consumptive-research question — a different tool and a different access path entirely.
Eligibility and Getting Access
Access to HTRC’s tools has generally been organized around institutional affiliation: researchers create an HTRC account, and Data Capsule access to in-copyright secure-mode text has involved additional steps beyond the account itself (algorithm review before non-consumptive outputs can be exported, in line with the non-consumptive framework above). The Extracted Features Dataset, being non-consumptive by design at the point of release rather than by access control, has been the more openly available of the two — downloadable without the same capsule-provisioning step. Because eligibility and account mechanics are the kind of detail that changes as infrastructure changes (and, per the next section, HTRC’s infrastructure is changing materially in 2026), confirm current account requirements directly on HTRC Analytics before planning a project around a specific access path.
Important: HTRC Services Are Discontinuing in Late 2026
This is a live, time-sensitive fact any research administrator advising on a project should know before committing to HTRC infrastructure. In October 2024, HathiTrust’s Executive Director announced that HathiTrust would cease funding and data access for the HathiTrust Research Center, with HTRC’s live services — including Data Capsule and web-based non-consumptive analysis tools that depend on current HathiTrust data — discontinuing after September 30, 2026. Reporting on the transition indicates that resources built on static, already-derived HTRC data (such as Bookworm) are expected to continue after that date, while services requiring ongoing, current access to HathiTrust’s live corpus (Data Capsule provisioning chief among them) will not. What happens to specific datasets and tools longer-term — retirement, integration into the general HathiTrust Digital Library, or continuation in some other form — was, per HathiTrust’s own transition communications, still being determined at the time of the announcement.
For a research administrator scoping a text-and-data-mining project that assumed ongoing HTRC access, this is a real planning constraint, not a hypothetical one: confirm current status directly against HathiTrust’s HTRC transition FAQ before scoping any new project around Data Capsule access, and treat any 2026-dated guidance elsewhere (including this page) as needing a fresh check against that FAQ rather than being taken as evergreen.
Where This Fits in a Research Data Management Workflow
For a research administrator supporting a digital-humanities or computational-text project, HathiTrust/HTRC access decisions intersect with several standard research data management concerns: whether derived outputs (word-frequency data, trained models) count as new research data requiring their own management and sharing plan; how non-consumptive access constraints interact with a funder’s data-sharing expectations, since the raw corpus itself can’t simply be redeposited in an open data repository the way originally-collected data can; and how text-and-data-mining copyright exceptions more broadly bear on what a project can legally do with a large text corpus, in-copyright or not. CASRAI’s guide to AI training data provenance, copyright, and TDM exceptions for research covers that broader copyright landscape in more depth; the FAIR Data Principles are also relevant background for thinking about how any derived, shareable outputs from an HTRC project should be described and released.
Frequently Asked Questions
Is HathiTrust the same thing as the HathiTrust Research Center?
No. HathiTrust is the digital library — the searchable catalog and reading platform for digitized volumes contributed by member libraries. HTRC is a separate research center, jointly run by Indiana University and the University of Illinois in partnership with HathiTrust, that builds the computational tools for non-consumptive text and data mining across that same corpus. Most general searches for “HathiTrust” are looking for the library catalog, not HTRC.
Can I read the full text of in-copyright books through HTRC?
No — that is precisely what the non-consumptive framework exists to prevent. HTRC tools let a researcher run computational analysis against in-copyright full text (inside a Data Capsule, for example) but do not let raw sequential text be exported or read as a substitute for the original work.
Do I need to be at a HathiTrust member institution to use these tools?
Historically, access has been organized around institutional affiliation and an HTRC account, with additional review for Data Capsule access to in-copyright secure-mode text. Confirm current eligibility directly with HTRC Analytics, especially given the 2026 service changes described above.
What’s the difference between the Data Capsule and the Extracted Features Dataset?
The Data Capsule is a secure virtual environment for running custom algorithms against full OCR text, including in-copyright volumes, with only derived outputs allowed to leave. The Extracted Features Dataset is a pre-built, freely downloadable set of page-level statistical features for the corpus — no capsule or custom algorithm required, but also no access to running text itself.
Is HTRC shutting down?
HathiTrust announced in October 2024 that it would stop funding and providing data access to HTRC, with live services discontinuing after September 30, 2026. Some HTRC-derived static tools and datasets are expected to continue past that date; confirm current status against HathiTrust’s own transition FAQ before relying on HTRC infrastructure for a new project.
