DiSSCo's Vision Applied: (Re-)connecting Fragmented Specimen Data through FAIR Digital Objects

In recent years, significant progress has been made in digitizing natural history collections using increasingly industrialized workflows involving conveyor belts, digital camera setups, robotics and Artificial Intelligence (AI). Also, new technologies became available to analyse the specimens. Analysis of both biodiversity and geodiversity samples has shifted from destructive analysis to non-destructive, high-resolution, and automated techniques accelerating the creation of new information.However, the resulting data is often fragmented across systems and repositories. Efforts to reconnect these data to the original specimen or derived samples frequently fail because identifiers were missing at the time of analysis, are not globally unique, change over time, or are referenced incorrectly. These issues can be solved by maintaining a digital object on the internet that is created at the time of collecting the sample, which contains contextual information and (links to) its derived data as this becomes available. This is called a Digital Specimen and different entities(human or machine) who create an analysis can add information to the digital object. A one-to-one relationship between the physical sample preserved as a specimen can be kept by giving the physical objecta persistent identifier like an IGSN, International Generic Sample Number. The digital object also gets a persistent identifier: a Digital Specimen identifier in the form of a FAIR Digital Object compliant DOI (Digital Object Identifier).

The Digital Specimen is a citable, machine-actionable proxy for physical specimens that is FAIR by design (FAIR Digital Object compliant) and has a Persistent Identifier (PID) in the form of a DOI to create a self-contained unit of knowledge. This design enables seamless linkage to derived data—such as chemical analysis, digital media, and publications. To implement this, DiSSCo (Distributed System of Scientific Collections) developed the open Digital Specimen (openDS) specification. By integrating community standards like Darwin Core with W3C PROV-O and JSON-LD, openDS provides a common semantic language for global interoperability.

DiSSCo is currently in transition from its project phase into becoming an operational European Research infrastructure. It has already created the first millions of FDO-compliant Digital Specimens and has developed infrastructure to allow the annotation of these digital objects with new data or improvements, either by humans or machines. AI fueled Machine Annotation Services (MAS) developed by third parties can operate in the infrastructure for analysis of the data or knowledge extraction from specimen images. 

In the presentation we will show how the FDO design supports advanced capabilities like multiple redirect to different digital representations for either human or machine, versioning and provenance to allow mutable objects, tooltips in journal systems that show contextual information about a referred sample in a publication through the PID record, and machine actionable metadata that supports machines to act on the data.