Changes from TCSv1.0 to TCSv1.01
At the Taxon Names/Concept sub-group meeting in St Petersburg and through discussion on TCS-LC Wiki following the meeting several issues were discussed for possible incorporation into the Release version of TCS - v1.01. The issues were considered to be minor if the change to the schema itself was minor e.g. an element name change or if there was consnesus on the list as to how a particular (usually taxonomic) issue should be represented. Issues which could have considerable implications that couldn't be tested in the time available or where there was no consensus amongst the taxonomists were deemed major issues which will form the basis for ongoing discussion. Future changes will be expected to be supported by evidence of rigorous testing with real data sets.
Changes
- Name changes - all elements have been made to have more specific names. e.g. 'is child of' -> 'is child taxon of', 'is parent of' -> 'is parent taxon of', Name -> AgentName, Relationships element -> TaxonRelationships, RelationshipAssertions -> TaxonRelationshipAssertions, Vouchers -> Specimens, Voucher -> Specimen, Specimen -> SpecimenItem
- Addition of role attribute to AgentName in order to handle 'ex' authors. Care should be taken when using this to ensure the correct semantics are captured and interpretted.
- TaxonName (formerly NameObject) has been changed to be an instance of ScientificName rather than extend it and an id attribute added to ScientificName.
- Minor changes to schema annotation to improve clarity.
Minor issues - unchanged
- 'is parent taxon of' to be used when modelling a taxonomic hierarchy as defined by a particular author and 'includes' to be used when representing a set based relationship between two TaxonConcepts. (similar for 'is child taxon of' and 'included in'). Therefore the order of relationships issues - isn't an issue they have different semantics.
- Vernacular relationships remain unchanged. There has not been sufficient investigation into the best way to model vernaculars. the existing method allows modelling but we need to have a sub-group investigate this more fully.
IsAnamorphic and nomenclaturalCode were proposed to be elements rather than attributes but we can't see the benefit or rationale for that - so it remains.
- The question was what the rational of treating some data attributes, others elements is. E.g. if simple types where to be treated as attributes, rank should be attribute as well.
xs:gYear doesn't cause any problems with simple integers.
- Not a problem, but probably unneccessary complexity introduced. Everybody now has to support 2002+05:00 or 2002Z in addition to 2002. Timezone is optional, but needs to be supported for xs:gYear.
It was suggested that the enumeration of types (for type specimens) was incomplete e.g no neo, lecto or epi however this is not the case they are all present.
- I cannot remember anyone suggesting this. However, I noted that the neo/epi etc. typication events within nomenclatural objects are not covered. Only Lectotypification seems to be covered.