The COAR Next Generation Repositories (NGR) initiative is a working-group-led effort, coordinated by the Confederation of Open Access Repositories (COAR), to define what an open-access repository needs to do — not just how it needs to store files — if it is going to be a full, interoperable participant in the scholarly communication ecosystem rather than a passive archive of deposited copies. It ran as a formal COAR working group from 2016 through the publication of its final report in November 2017, and its recommendations continue to shape how repository platforms, and the standards work built on top of them, are designed.
Why COAR launched it
By the mid-2010s, thousands of institutional and disciplinary repositories were running worldwide, most built around OAI-PMH metadata harvesting and structured primarily to support green open-access deposit — making a copy of a publication (or, less often, a dataset) discoverable and downloadable. COAR’s own assessment, reflected in the working group’s charge, was that this first generation of repository infrastructure had succeeded at that narrow job but had not kept pace with what the broader web-native scholarly ecosystem had come to expect: machine-to-machine interoperability, participation in peer review and evaluation workflows, support for text and data mining, and integration with the growing web of persistent identifiers connecting people, organizations, and outputs. Repositories risked becoming isolated silos next to increasingly connected commercial and infrastructure services. NGR was COAR’s response: a coordinated attempt to define a shared technical direction so that repositories, collectively, could close that gap rather than each platform solving it in isolation.
Origins and timeline
COAR launched the Next Generation Repositories Working Group in April 2016 to identify the core functionalities the next generation of repositories would need, along with the architectures and technologies required to implement them. The working group ran an open, community-input process: in February 2017 it published draft user stories for public review and comment, then used that feedback to converge on a defined set of repository behaviours. The working group’s final report — Next Generation Repositories: Behaviours and Technical Recommendations of the COAR Next Generation Repositories Working Group, authored by Andrea Bollini, Peter Knoth, Pandelis Perakakis, Eloy Rodrigues, Kathleen Shearer, Herbert Van de Sompel, and Paul Walk — was published in November 2017 (version 2, dated November 28, 2017).
What the report actually defines
The report specifies 11 behaviours a next-generation repository should support, together with a list of roughly 19 recommended technologies, standards, and protocols repository platforms could adopt to implement them. Rather than prescribing a single piece of software, the report deliberately frames these as behaviours and technical building blocks — the outcomes a repository should be capable of, and the standards-based means to get there — so that different repository platforms (institutional, disciplinary, commercial, or open source) could each implement them in their own way while remaining interoperable with one another.
Broadly, the behaviours cluster around a few functional themes that run through the whole report:
- Discovery and access — making deposited content, and the metadata describing it, genuinely findable and retrievable by both humans and machines, not just harvestable via a single legacy protocol.
- Reuse and text/data mining — exposing content in forms (formats, licensing signals, machine-readable metadata) that support computational reuse, not only human reading.
- Interoperability with evaluation and peer review — allowing repositories to participate in review, annotation, and assessment workflows run by services layered on top of the repository, rather than treating peer review as something that only happens before deposit, inside a journal’s own system.
- Notification and awareness — giving repositories a standard way to notify, and be notified by, other services in the ecosystem when something relevant happens to an item (a review is posted, a citation is made, a related version appears elsewhere).
- Identification — consistent use of persistent identifiers for people, organizations, and outputs, so a repository record can be reliably linked into the wider research-information graph rather than existing as an isolated page.
This framing is why NGR sits squarely at the intersection of repository infrastructure and the identifier/interoperability standards CASRAI’s persistent identifiers and research information systems content covers: the report is, in effect, a statement of what a repository needs to interoperate correctly with ORCID, DataCite, ROR, and CRIS platforms, rather than a general repository-software feature list.
From behaviours to protocol: COAR Notify
NGR’s report was a statement of direction, not itself a protocol implementation — and one of the clearest lines from the report into working infrastructure is the COAR Notify Initiative. COAR Notify operationalizes the “notification and awareness” behaviour by defining a set of profiles, constraints, and conventions around the W3C Linked Data Notifications (LDN) recommendation, so that repositories and external services (peer-review overlay platforms, annotation tools, evaluation services) can exchange structured, machine-readable notifications about a given research output in a distributed, web-native way rather than through bespoke point-to-point integrations. The initiative has been funded as a multi-year project by Arcadia, and its ecosystem of participating repository platforms and services has continued to expand well past NGR’s original 2017 report.
2020: a common architecture for distributed peer review
NGR’s influence continued past the 2017 report. In January 2020, COAR convened a meeting hosted by Inria in Paris to investigate whether a common, distributed architecture could connect peer review directly to resources held in repositories — effectively repository-hosted overlay review, rather than review happening only inside a journal’s own submission system. That work led to a proposed model for a standardized approach to peer review of repository resources, which COAR released for public comment in August 2020. Both efforts are direct extensions of the “interoperability with evaluation and peer review” behaviour identified in the original NGR report, and both fed into what became the COAR Notify Initiative’s protocol work.
Why this matters for research administrators and CRIS/RIM teams
Most research administrators will never read the NGR report directly, but its consequences show up in day-to-day infrastructure decisions:
- Repository procurement and platform evaluation. When an institution evaluates or upgrades its repository platform, NGR-derived capabilities — identifier-based linking, notification support, machine-readable exposure of content — are a reasonable checklist item precisely because they map onto interoperability with the ORCID, DataCite, ROR, and Crossref infrastructure most institutions already depend on for reporting.
- CRIS interoperability. A repository that implements NGR-style behaviours is easier to connect to a CRIS: identifier-consistent metadata and standard notification patterns reduce the custom integration work otherwise needed to keep a CRIS and a repository’s holdings in sync.
- Open-access and open-science compliance. Funder and institutional open-access mandates increasingly expect not just deposit, but deposit that is discoverable, linked to persistent identifiers, and machine-actionable for compliance reporting — exactly the outcomes NGR’s behaviours target.
- Vendor conversations. Citing COAR’s NGR recommendations and the COAR Notify protocol gives research administrators a standards-based reference point when asking a repository vendor or open-source platform community whether, and how, a given interoperability feature is supported.
Current status
The COAR Next Generation Repositories Working Group’s formal deliverable is the 2017 report; COAR has not issued a comparable numbered successor report specifically under the “Next Generation Repositories” name. Its substance continues to develop primarily through the COAR Notify Initiative, which remains an active, expanding project connecting repositories to review and evaluation services on the NGR-defined interoperability model, and through COAR’s ongoing advocacy and technical work with repository networks and platform communities worldwide.
Frequently asked questions
What is COAR?
COAR (the Confederation of Open Access Repositories) is an international association of libraries, universities, research institutions, and other organizations that operate or support open-access repositories. It was founded in 2009 to strengthen the global network of repositories and advance common approaches to open access and repository interoperability, and today counts well over 150 member and partner organizations worldwide.
Is Next Generation Repositories a piece of software?
No. NGR is a set of behaviours and technical recommendations — a direction for repository platforms to build toward — not a single software product. Individual repository platforms (open source and commercial) each decide independently how, and how fully, to implement the recommended behaviours and standards.
How is NGR different from COAR Notify?
NGR is the earlier, broader initiative that defined the target behaviours (2016–2017); COAR Notify is a specific, ongoing protocol project that implements one of those behaviours — standardized, repository-to-service notification — using the W3C Linked Data Notifications specification.
Does NGR relate to persistent identifiers like ORCID, DOI, and ROR?
Yes. Consistent use of persistent identifiers for people, organizations, and outputs is one of the functional themes running through the NGR report, since a repository can only participate in the kind of interoperable, machine-readable ecosystem NGR describes if its records reliably link to identifiers like ORCID iDs, DOIs, and ROR IDs rather than relying on free-text names.
