Difference between revisions of "Audubon Core Structure"

Revision as of 09:12, 11 October 2012

Title: Audubon Core

Date: TBD. This document is a proposal

Abstract: The Audubon Core is a set of vocabularies designed to represent metadata for biodiversity multimedia resources and collections. These vocabularies aim to represent information that will help to determine whether a particular resource or collection will be fit for some particular biodiversity science application before acquiring the media. Among others, the vocabularies address such concerns as the management of the media and collections, descriptions of their content, their taxonomic, geographic, and temporal coverage, and the appropriate ways to retrieve, attribute and reproduce them. This document contains material introductory to the Audubon Core Term List

Contributors: Robert A. Morris, Vijay Barve, Mihail Carausu, Vishwas Chavan, Jose Cuadra, Chris Freeland, Gregor Hagedorn, Patrick Leary, Dimitry Mozzherin, Annette Olson, Greg Riccardi, Ivan Teage

Legal: This document is governed by the standard legal, copyright, licensing provisions and disclaimers issued by the Taxonomic Databases Working Group.

Part of TDWG Standard: TBD

This document has revision ID 5404 which can be permanently accessed through http://terms.gbif.org/w/index.php?oldid=5404.
The version under consideration by TDWG has permalink http://terms.gbif.org/w/index.php?oldid=5396.

ONCE REVIEW AND REVISION FINALIZED: This is a normative document.

This is the normative documentation for the TDWG Audubon Core Multimedia Resources Metadata Standard (Audubon Core, or simply AC), During development, it was colloquially known as MRTG, after its developers, the GBIF-TDWG Joint Multimedia Resources Metadata Task Group. Please see the brief Audubon Core Non Normative Document and also MRTG Development History for the development history in detail.

If you are unfamiliar with the AudubonCore, please read the Audubon Core Non Normative Document before editing this page. It lays out why there is perceived a need for a biodiversity media resource metadata schema, and how the standard attempts to use existing metadata standards where possible.

Related Information

• See also discussion of theXML Schema For MRTG. This is obsolete and will be revised for compliance with the normative version when it is approved.
• See also MRTG in RDF. This is simple RDF is generated automatically from the normative Term List, and will generally be current.
• TDWG09 MRTG WORKGROUP REPORT

Terminology of this specification

There are many ways to organize metadata specifications, particularly as to the nomenclature of the constituents of the metadata. In this document and the associated non-normative documentation, we will follow closely (sometimes verbatim) a portion of the Dublin Core Metadata Initiative (DCMI) metadata nomenclature as described in Section 2.3 of the DCMI Abstract Model (http://www.dublincore.org/documents/abstract-model/). In addition:

• A Multimedia Resource is anything that a provider identifies as belonging to one of the possible values of the AC Type term and one of the Subtype term values. A mechanism is provided by which providers can supply a privately defined subtype that will not collide with the AC defined Subtype values.
• An AC record is a set of terms with any values conforming to this document, and which contain at least the four mandatory terms described in the Audubon_Core_Term_List, and which describes a single multimedia resource (possibly including a Collection). One of these, the value of Identifier is a Globally Unique IDentifier (GUID), which may have been assigned to the resource by an external authority or by the provider of the metadata record.
• AC terms are divided into two Layers. Those characterized as in the Core Layer, including the five mandatory terms, should be meaningfully handled by all consuming clients applications. Only wholly complete consuming applications need handle those in the Extended Layer. What is meant by "meaningfully handle" is up to implementers of this normative specification. It could be as simple as "gracefully ignore".

In the Audubon_Core_Term_List, every AC term has a term name following a table entry "Term:", a URI, a plain text normative Definition, a recommended English Label, an optional Details attribute, and an optional Comments attribute. In addition, a term has an attribute telling whether it is mandatory, one telling whether it is repeatable, and one telling whether it is in the Core or Extended Layer. The Extended Layer comprises terms likely to only occur for certain media. For example, the term DateAvailable will apply only to media that are embargoed, but for which the provider is prepared to make the metadata immediately available.

AC metadata can describe either individual multimedia resources or collections of resources. A few, but not many, of the AC properties have different values for collections than for individual media. If no such distinction is mentioned, AC does not assume one.

Term Names for terms borrowed from other vocabularies are those in use for the corresponding term in those vocabularies. Term names are intended principally for navigation in this document. Term Labels are suggestions for English labels in applications. They are recommendations only and are offered only in English, with the added expectation that they may clarify intended usage of the term. Communities may wish to promulgate recommendations for Labels in other languages, or even alternative English Labels for specialized audiences, e.g. school children.

URI's for terms conform to the http URI scheme (See, http://en.wikipedia.org/wiki/URI_scheme, http://www.w3.org/TR/uri-clarification/, or http://www.ietf.org/rfc/rfc2396.txt ). Informally, one may understand this as follows: an http URI has the syntax of an http URL, but there is no expectation that putting it in a web browser will result in any information being returned to the browser, and if there is, it may have no relevance. This conformance requirement applies only to the URIs that identify AC terms. A few AC terms permit values to be taken from another controlled vocabulary chosen by the user. In this case, those values may involve URIs conforming to a scheme given by that external vocabulary, and AC is silent on what that scheme is.

The Details field of a term's documentation points to further information, if any exists, about the term. In particular, for terms borrowed from other vocabularies, this field generally carries a link to the originating vocabulary's documentation for that term.

Multiplicity/Cardinality

A number of terms are repeatable. How to implement repeatability in a given serialization is not defined by Audubon Core. The following section gives advice on some best practices in the context.

The simplest case is a single repeatable term (e.g., dcterms:identifier). In xml schemata permitting elements to be repeated (e.g., RDF/XML) such a term may simply be repeated (e.g. "...<dcterms:identifier>http://example.com/123</dcterms:identifier><dcterms:identifier>http://example.com/456</dcterms:identifier>..."). In serializations that do not easily lend themselves to repeatable elements (e.g. "flat" schemata with all elements occurring only a single time in an otherwise unstructured record) it is possible to define separators to support a list of values within a single element (e.g. "...<dcterms:identifier>http://example.com/123; http://example.com/456</dcterms:identifier>...").

In certain cases pairs or tupels of properties are repeated. In Audubon Core this situation occurs, for example, in the following cases:

• The language-dependent metadata like title, description, etc. need to be associated with ac:metadataLanguage. One approach here is to use complete Audubon Core records together with the ac:metadataLanguage property, see there for further detail.
• The values of properties about a Service Access Point must remain associated with that Service Access Point even if there are multiple Service Access Points. See hasServiceAccessPoint for further details.
• The terms dwc:scientificName and dwc:identificationQualifier may optionally be structured into pairs (see the notes on identificationQualifier).
• The terms Reviewer, being the name of an individual providing some expert review of a resource, and the review text itself in Reviewer Comments are desirable to store as pairs.

Many serialization languages provide sufficiently structured forms. A first example in XML using a nesting structure might be something like this:

<MEDIA_METADATA_CONTAINER>
  <dcterms:identifier>http//:example.com/pictures/thePicture.jpg</dcterms:identifier>
  ...
  <ac:hasServiceAccessPoint>
    <dcterms:format>jpg</dcterms:format>
    <ac:accessURI>http://example.com/fullres/thePicture.jpg</ac:accessURI>
    ...
  </ac:hasServiceAccessPoint>
  <ac:hasServiceAccessPoint>
    ...
  </ac:hasServiceAccessPoint>
<MEDIA_METADATA_CONTAINER>

Another example may reference access points by identifier:

<MEDIA_METADATA_CONTAINER>
  <dcterms:identifier>http//:example.com/pictures/thePicture.jpg</dcterms:identifier>
  ...
  <ac:hasServiceAccessPoint>http//:example.com/pictures/thePicture.jpg#ac0001</ac:hasServiceAccessPoint>
  <ac:hasServiceAccessPoint>http//:example.com/pictures/thePicture.jpg#ac0002</ac:hasServiceAccessPoint>
  <ac-classes:ServiceAccessPoint id="http//:example.com/pictures/thePicture.jpg#ac0001">
    <dcterms:format>jpg</dcterms:format>
    <ac:accessURI>http://example.com/fullres/thePicture.jpg</ac:accessURI>
    ...
  </ac-classes:ServiceAccessPoint>
  ...
<MEDIA_METADATA_CONTAINER>

Note: ac-classes:ServiceAccessPoint is used, e.g., in Audubon_Core_Term_List_RDF_Version but not formally defined or required by AC.

Where such structure is impossible or undesirable, an alternative solution is to to permit only one access point per MEDIA_METADATA_CONTAINER, but to repeat the main MEDIA_METADATA_CONTAINER for a single media resource. This is similar to one of the options discussed for multilingual metadata (see ac:metadataLanguage). An example in XML for this:

<MEDIA_METADATA_CONTAINER>
  <dcterms:identifier>http//:example.com/pictures/thePicture.jpg</dcterms:identifier>
  <dcterms:title>A red beech leaf</dcterms:title>
  <dcterms:format>jpg</dcterms:format>
  <ac:accessURI>http://example.com/fullres/thePicture.jpg</ac:accessURI>
  ...
<MEDIA_METADATA_CONTAINER>
<MEDIA_METADATA_CONTAINER>
  <dcterms:identifier>http//:example.com/pictures/thePicture.jpg</dcterms:identifier>
  <dcterms:format>png</dcterms:format>
  <ac:accessURI>http://example.com/fullres/thePicture-hires.png</ac:accessURI>
  ...
<MEDIA_METADATA_CONTAINER>

The same example as a spreadsheet-like table:

In the example above, only the required identifier is repeated, but not the title field. Whether to repeat all fields or whether to provide all fields only in the first record, limiting later records to the identifier and the service access point properties, is left to specific implementations.

Lists of plain text values

Some AC terms permit values that are lists to be represented as plain text. The choice of how to separate list items is ultimately left to the implementers of AC. Typical usage is to choose a punctuation mark such as ",", ";", or "|". In these cases a special escape syntax needs to be defined for cases in which the separator is part of the metadata value. Unfortunately, even for standard list formats like CSV, different software packages choose different escape methods, hindering. In the absence of an implementation-specific choice we recommend to use "|" as separator and "\|" as an escaped vertical bar.

Term List

See: Audubon Core Term List (1.0 normative)

Non-normative documents

See: Audubon Core Non Normative Document