ObjectOntology

GregorHagedorn - Tue Mar 15 2005 - Version 1.5
Parent topic: WebHome
This topic is about attempting to subdivide the totality of biodiversity informatics into separate knowledge domains, for which object types may be defined. Bioidiversity data could then be expressed in a matrix (or composition) or such object types, see DataModelMatrix. Ultimately, the object should be in a - relatively flat - ontological hierarchy, so any suggestions about this are appreciated.

Below a rough first draft based on SDD/UBIF. We should work towards a rough consensus on how to delimit knowledge domains, and which labels for the object types are considered most useful and intuitive. In SDD we have:

• DescriptiveData (consisting of DescriptiveTerminology, CodedDescriptions, NaturalLanguageDescriptions, and (fixed) IdentificationKeys)
  plus inherited from UBIF (currently as external interface or "proxy" objects, to be revised as outlined in DataModelMatrix):
• ClassNames (= definition of vernacular or scientific names)
• ClassHierarchies (= taxonomic hierarchy, including synonmyization)
• Objects (ABCD Units, ObjectOccurrences?, TaxonOccurrences?

or Specimens? Problem: some uses need to validate that only stored/preserved objects are adderessed (e.g. nomenclature), others generalize stored and observed objects, some uses apply to parts of objects)

• Agents (person, institution, software)
• Publications (references to digital/printed publications

or online-publications itself)

• Geography (geographical locations and areas.

Problem: objects in collection are called Location rather than Geography again. Alternative Names: Locations/Location (not clear whether geographic or in book, document, etc.) GeoLocations/GeoLocation - perhaps best choice?

• MediaResources (image, audio, video, formatted text like pdf, html, etc.)
• MeasurementUnits (scientific like m, kg, s, non-scientific like in, ft, oz)

In the species bank workshop in Amsterdam, I further proposed:

• Uses (biotechnological, medical, agricultural and mythological uses of plants)
• Conservation (in the sense of nature conservation/red list data, would be good to have better term, conservation may also be applied to dead objects to conserve against decay)
• Ecology - large area, no details ontology from me, except for
  • OrganismInteractions = two organisms, an interaction type like host-pathogen, predator-prey, or pollinator, and area in which this interaction is known to occur
• Distribution in the sense of chorology, i.e. synthetic/summarized hypothesis about actual or potential distribution.

Additional types in the overall object concept, for which no root collections would be defined because they are only used in compositions may be NomenclaturalOpinion, Identification, and several types like character, concept, modifier used inside SDD.

Which other types (perhaps as subtypes in an ontology) should be defined for a start? Which names you simply dislike, even if you don't have a better one?

In email to tcs-lc list, Roger Hyam proposed a root collection "Nomenclature". This corresponds with Geography, which also differs from the plural form of the class name itself. In general I think it desirable to stay with the "plural name for collections" pattern, but intuitive names may be even more important.

-- Main.GregorHagedorn - 09 Mar 2005

The ontology I use in The Taxonomicon looks like this:

• Nomenclature
  • Rank (item)
  • Scientific name (text)
  • Common name (text)
  • Epithet (text)
  • Author citation (text)
  • Nomenclatural attribute (item)
  • Nomenclatural code (item)
• Taxonomy
  • Taxon group (item)
  • Taxon (taxon)
• Geography
  • Biogeographic realm (item)
  • Geographic region (item)
  • Type locality (text)
  • Presence (item)
• Geology
  • Geologic time (item range)
  • Geologic age (number range)
  • Year of extinction (number)
• Ecology
  • Habitat
    • Ecologic realm (item)
    • Major habitat type (item)
    • Ecoregion (item)
    • Habitat (item)
  • Relationships
    • Host (taxon)
    • Host type (item)
    • Type host (boolean)
    • Parasite (taxon)
    • Parasite type (item)
    • Pathogen (boolean)
    • Disease (text)
    • Symbiont (taxon)
    • Symbiont type (item)
    • Commensal (taxon)
    • Commensal type (item)
    • Host dependency (item)

GBIF should provide a registry for all standardized objectID's, data source names and parameter names.

-- Sheila Brands, Universal Taxonomic Services, The Taxonomicon & Systema Naturae 2000 - 10 Mar 2005

(Donald Hobern and Sheila Brands propose a closely related discussion on whether an object-oriented approach is appropriate at all. Please discuss this under KeywordBasedDataExchange! -- Gregor - 15. March 2005)