Explainer · Plain-language

The Fair Digital Object: Definition, Meaning & Examples | CASRAI

A FAIR Digital Object (FDO) is a digital entity — data, software, a model, or any other research output — that has been made machine-actionable by combining it with a persistent identifier, a type record describing what it is and how to interact with it, and structured metadata conforming to agreed standards. The FDO concept extends the FAIR principles by defining a concrete technical framework for making research objects not just findable and accessible to humans, but actionable by machines and services.

CASRAI plain-language explainers — clear answers to recurring research-administration questions

The step most authors miss

Doing CRediT right? Don’t stop at the statement.

A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.

Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.

Origins and the FDO Forum

The digital object framework was conceived by Robert Kahn and Robert Wilensky at CNRI (Corporation for National Research Initiatives) in the 1990s, culminating in their 2006 paper in the International Journal on Digital Libraries. CNRI developed the Handle System — the persistent identifier infrastructure on which DOIs are built — as part of this framework. The vision was of a distributed digital object infrastructure in which objects could be identified, located, accessed, and operated on programmatically regardless of where they were stored. The FDO Forum was established in 2019 as an international body to develop FDO specifications, governance frameworks, and community adoption strategies. It brings together research infrastructures, data repositories, and e-infrastructure providers from Europe, North America, and beyond. The Research Data Alliance (RDA) has active working groups on FDOs that produce recommendations feeding into the FDO Forum's specification work. EOSC has run FDO pilots as part of its strategy to make European research data machine-actionable at scale.

Components: PID, type record, and DOIP

An FDO is defined by three layers working together. The **persistent identifier (PID)** provides a stable, globally unique reference to the object, independent of its physical location. Handle System identifiers (including DOIs) are the most common PIDs used in FDO implementations. Unlike a simple DOI landing page (which resolves to a human-readable web page), an FDO PID resolves to machine-readable metadata that enables programmatic operations. The **type record** (FDO profile) is a machine-readable schema that defines what the FDO is (its type), what metadata it carries, and what operations can be performed on it — for example, retrieve, update, or validate. Type records are themselves FDOs stored in a type registry. The **Digital Object Interface Protocol (DOIP)** is the communication protocol developed by CNRI for performing operations on digital objects. It defines standard operations (Create, Retrieve, Update, Delete) and allows custom operations defined by the type record. Cordra, an open-source software package developed by CNRI, provides a reference implementation of DOIP and the FDO framework.

FDO vs FAIR data and Linked Data

FAIR data principles (Wilkinson et al., 2016, Scientific Data) describe the properties that digital objects and their metadata should have to be reusable. FDOs provide a technical architecture for achieving FAIR properties, particularly the Interoperable and Reusable dimensions. An FDO is explicitly machine-actionable: not just described with metadata that humans can read, but operable by services that can discover what operations are possible and execute them without human intervention. Linked Data (RDF, JSON-LD, schema.org) provides an alternative and complementary approach to machine-actionable research objects, using web standards and HTTP URIs rather than the Handle/DOIP stack. Both approaches make digital objects machine-readable; the key differences lie in protocol governance (W3C standards for Linked Data, FDO Forum specifications for FDOs), adoption community, and tooling maturity. RDF-based Linked Data has deeper adoption in the library and semantic web community; FDOs have stronger traction in large European research infrastructure projects. The two approaches are increasingly seen as complementary rather than competing.

Key facts

At a glance

FDO concept: Kahn and Wilensky (2006), International Journal on Digital Libraries, doi:10.1007/s00799-005-0128-x
FDO Forum established 2019 to develop FDO specifications and governance
Three FDO components: persistent identifier (PID), type record (FDO profile), and bit sequences (data/metadata)
DOIP (Digital Object Interface Protocol): the communication protocol for FDO operations, developed by CNRI
Cordra: open-source CNRI reference implementation of the FDO/DOIP framework
EOSC FDO pilots: part of the European Open Science Cloud strategy for machine-actionable research outputs

Common misconceptions

What people often get wrong

Often heard: A FAIR Digital Object is just a dataset with a DOI.

Actually: A DOI alone makes an object findable and provides a stable link — but a DOI landing page is typically a human-readable HTML page. An FDO goes further: the PID resolves to machine-readable metadata (a type record/FDO profile) that allows software agents to discover what the object is, what operations it supports, and how to interact with it programmatically without human intervention.

Often heard: FDOs replace the FAIR principles.

Actually: FDOs are a technical implementation framework designed to realise the FAIR principles, not a replacement for them. The FAIR principles describe properties data should have; the FDO framework provides one concrete technical pathway — using PIDs, type records, and DOIP — to achieve those properties in a machine-actionable way.

Often heard: FDOs and Linked Data are competing approaches that cannot coexist.

Actually: Both FDOs and Linked Data aim to make digital objects machine-readable and interoperable. They use different protocol stacks (DOIP vs HTTP/RDF), but they are increasingly treated as complementary. Many FDO implementations incorporate RDF metadata, and research data infrastructures often expose both DOIP and Linked Data access points.

Going deeper