Welcome

Thank you for your interest in offering a Machine Annotation Service (MAS) for the DiSSCo community! Here you’ll find information on how to adapt existing services for the DiSSCo platform with minimal effort.

About DiSSCo

DiSSCo is a research infrastructure supporting digitized natural science collections. With over 200 partners in 23 countries, DiSSCo aims to digitally unify all European natural science assets. It provides digital specimens and digital media objects from diverse collections in a single harmonised data model, openDS. These harmonised data are made available through a user-friendly platform, DiSSCover, or through a REST API.

The DiSSCo core architecture aims to go beyond harmonising the data, however. Data within DiSSCo can be evaluated, enriched, and extended by the community, improving data quality across institutional boundaries. The way DiSSCo achieves this is through annotations.

What is an Annotation?

An annotation is an additional piece of information associated with a specimen or media object. An annotation is initially separate from its target. The annotation is then evaluated by experts, who may reject or accept the annotation. An accepted annotation may then change the specimen or media object in the source system. This approach allows for widespread community curation of the entire digitized European collection.

Also see the annotation JSON Schema

What is a Machine Annotation Service?

A Machine Annotation Service (MAS) is an automated service that annotates a target in DiSSCo. These services are scheduled by users on individual specimen or media in DiSSCo. What a MAS offers is broad. From sophisticated AI services to taxonomic services to linking to other infrastructures, MASs add value to natural science collections data in all sorts of ways.

What can be annotated?

In DiSSCo, two kinds of objects can be annotated: digital specimens or media objects. The specific object being annotated is the “target” of the annotation.

What part of the target is being annotated is another important aspect of the annotation. The annotation may be as broad, or as narrow, as necessary. This is called the “selector”, and relates specifically to the data according to the OpenDS specification.

  • Whole object: The whole specimen or media is being annotated
  • Class: A whole class (e.g. TaxonIdentification or Event) is being annotated
  • Term: A specific, individual property (e.g. dwc:genus or dwc:locality)
  • Region of Interest: (Media only) A specific area of the image is being annotated

What terms can be annotated? The DiSSCo Terms Site has the most up to-date information on openDS terms.

Why make an annotation?

The “motivation” of an annotation, or the “why” of it all, is an important part of the annotation. It determines how the annotation may change its target. The possible motivations are:

  • Adding: The user wants to add new information to the specimen/media
  • Deleting: A piece of information is no longer relevant to a specimen/media, and should be removed
  • Assessing: The user wants to assess the quality of the data in the specimen/media (e.g. data quality flags)
  • Editing: The user wants to edit an existing value of the specimen/media
  • Commenting: The user wants to make a generic comment on the specimen/media

Note: Only the ods:adding may reference a part of the target that doesn’t exist yet because you are adding information to the target. If you are creating an annotation with another motivation, you must make sure you’re selecting on a part of the target that exists. (How can you comment or edit something that doesn’t exist?)

Who can make annotations?

Human experts can manually annotate resources using DiSSCo’s DiSSCover platform. However, machines may annotate on a much larger scale. Machine Annotation Services (MASs) are computational services that automatically produce annotations on digital objects. These services may be triggered automatically, when a digital object is first ingested into the system, or may be scheduled by a user in DiSSCover. MASs often use external APIs or AI to produce their annotations.

Why develop a MAS?

If you’ve developed a service that analyzes specimen data, you may be wondering why bother with publishing it on DiSSCo. Haven’t you done enough work?

  • One data model to rule them all: one service can be applied to data from hundreds of institutions
  • Modular design: The DiSSCo architecture is designed to allow existing services to “plug in” easily, meaning anyone, not just those directly tied to DiSSCo, can develop a MAS
  • Ease of use: Through the DiSSCover interface, even the least tech-savvy user can schedule sophisticated machine services on their data

These three factors mean that a single service can have a much wider reach through DiSSCo. Through collaboration, we can further our shared mission of fighting the biodiversity crisis.