UBIF as a framework for a biological object type model (G. Hagedorn & al.), proposing an ID system, simple text labels as lowest possible level of detail, plus core level and detail level. However, the UBIF proposal assumed that always one knowledge domain would be represented in detail (e.g. SDD) and the others would be present only in the form of "proxy objects" or "data interfaces" on the base or core level.
It emerged, that a framework and an ontology of biological object types was necessary, and that object types had to be defined at various level of detail. Thus DarwinCore and ABCD are not different solutions to the same problem, but complementary solutions that are needed in coexistence.
At the end of TDWG, Rebecca Shapley presented a vision of how future data sets may be a mosaic of object types and detail of representation (see http://www.tdwg.org/2004meet/EV/TDWG_2004_Papers_Shapley_1.zip). I have slightly reworked her slides and present them below as a background to the discussion of how the future object ontology and data framework might look:
Proposal for different level of detail for object representations. To those displayed, additional types should be added intended for markup of existing unstructured text (or low structured mixed content such as xhtml), similar to the NaturalLanguageDescription markup type in SDD.
Matrix formed by level of detail x different object types used in biodiversity informatics (= classes).
Examples how the focus of accepted or proposed TDWG standards maps unto the matrix. (##The title of the slide should be renamed from "Identity" to "Inventory" or "Focus of existing work")
Examples how the entire DarwinCore and ABCD might related to the matrix once reformulated as object types.
Examples how SDD relates to the matrix (SDD is already formulated through object types).
Levels of detail should have compatibility so that consumers may prefer to consumer data on a lower level.
---
Related to the above, especially the topic of reformulating complex and core standards as a composition of objects types, four slides from the UBIF talk held in Christchurch 2004. These contain an outline of how existing standards might be modularized by reusing the same object types in various compositions:
The last image shows the object types proposed in UBIF and how linking mechanism (bottom), a base representation as an unstructured text label in multiple languages (middle) and specialized interfaces (equivalent to the "core" style) might build on each other. For the matrix framework, this vision need to be extended to also layer complex object types on top.
An interesting aspect of viewing objects not as a singular reality, but as having multiple representations of different detail, is that the conventional distinction between the object and a reference to an object is blurred. See ObjectTypePattern for further discussion of this.
-- Main.GregorHagedorn & Rebecca Shapley - 09 Mar 2005
Perhaps then the columns in the framework are "Conceptual Elements", the cells (at the intersection of a column and a level) are "Concept Object Models", some of which will have colorful names (like AlexandriaCOM, LinneanCOM), and the combination of Concept Object Models into TDWG Standards for Data Exchange are Concept Object Compositions (DwC being an example). The framework/grid itself is a Data Exchange Interface. Interface in the sense that one programs against them.
Proposal for different level of detail for object representations. To those displayed, additional types should be added intended for markup of existing unstructured text (or low structured mixed content such as xhtml), similar to the 
Matrix formed by level of detail x different object types used in biodiversity informatics (= classes).
Examples how the focus of accepted or proposed TDWG standards maps unto the matrix. (##The title of the slide should be renamed from "Identity" to "Inventory" or "Focus of existing work")
Examples how the entire 
Examples how SDD relates to the matrix (SDD is already formulated through object types).
Levels of detail should have compatibility so that consumers may prefer to consumer data on a lower level.
---
Related to the above, especially the topic of reformulating complex and core standards as a composition of objects types, four slides from the UBIF talk held in Christchurch 2004. These contain an outline of how existing standards might be modularized by reusing the same object types in various compositions:
The last image shows the object types proposed in UBIF and how linking mechanism (bottom), a base representation as an unstructured text label in multiple languages (middle) and specialized interfaces (equivalent to the "core" style) might build on each other. For the matrix framework, this vision need to be extended to also layer complex object types on top.