In 2009, Darwin Core (DwC) was accepted as a TDWG standard and in 2011 the joint GBIF/TDWG Media Resources Task Group (MRTG) completed its work and has submitted its multimedia resources vocabulary, Audubon Core, for review as a TDWG standard. These two standards provide many of the metadata terms necessary to describe live plant images and the plants that they document. Wherever possible, Bioimages uses terms from these vocabularies to describe resources in the collection.
The Darwin Core TDWG standard contains many terms to describe data properties of biodiversity resources, such as information about the location and identification of the occurrence of an organism. It does not generally provide terms to describe the relationships among different types of resources (i.e. object properties). The Darwin-SW (DSW=Darwin Semantic Web) ontology expands upon the basic DwC vocabulary by formally defining resource classes and the relationships among them. Bioimages uses DSW to describe resources in Resource Description Framework (RDF).
The Multimedia Resources Task Group (MRTG) has established standardized terms for metadata used to describe all kinds of media, including images. The vocabulary (known as Audubon Core; AC) containing these terms has been submitted for review as a TDWG standard. Whenever appropriate, Bioimages uses AC to describe the media aspects of resources in the collection.
A structural feature of Audubon Core is that a particular media resource may be represented by several versions that are defined as instances of "service access points". Each service access point has a URL that can be used to retrieve that particular version and descriptive terms for the file type (e.g. jpeg, gif, tiff, etc.) and the resolution (high resolution, low quality, thumbnail, etc.). All of the service access points can share terms that describe metadata that they have in common (e.g. title, owner, credit line, etc.).
Bioimages has adopted a particular metadata organizational structure that allows for more complex relationships than can be represented in a typical "flat" database. The diagram above illustrates the relationships among resources that were created in a complex network of derivation (a botanical garden scenario). The most significant feature of this system is that a resource representing the individual living organism in its environment acts as a node which connects all occurrences derived from the individual as well as one or more taxonomic determinations. This system is described in detail in Baskauf (2010) and has been subsequently modified in the development of Darwin-SW.
The advantage of this structure in the case of plant imaging is that a single system can describe the relationships among specimens and their images (above left), and multiple images of living organisms (above right). The system can also handle both specimens and living organism images from the same individual (not shown).
Current linked data and GUID practices require that metadata for resources be described using Resource Description Framework (RDF). RDF describes the properties of a resource and specifies how the resource is related to other resources. RDF relationships can be expressed by "graphs" - the RDF graph of occurrence 2 from the first figure is shown below:
The arrows pointing to boxes represent properties of the occurrence that can be represented by string literals. The arrows pointing to ovals represent properties that indicate relationships with other resources. The properties are generally defined by terms from Darwin Core, Audubon Core, or terms (using the namespace dsw:) that can represent relationships needed by the organizational system described above. Click here to see a graph of an individual and here to see the graph of a determination.
Since computers can't read graphs, RDF is generally expressed in a form more easily consumed by a computer, such as XML. Click here to see the RDF for an individual and its determinations in XML format. Click here to see the RDF for an occurrence (live plant image and its service access points) in XML format.
RDF in XML is the consensus format in which metadata should be delivered when a GUID is resolved by a computer (as opposed to a human using a web browser).
