Technology/Approach to core Vocabulary/Ontology
Hi All,
I believe we agree that we need a core vocabulary/ontology and that this is limited in scope and under the remit of the TAG for the time being.
So my first question is: *How do we develop a core vocabulary/ontology in a distributed, collaborative way?*
Renato pointed out the way CIDOC is modeled which is similar (but more complex) to the way DublinCore is presented. Should we just start with a wiki page and build it from their or is there a more sophisticated tool we should use?
Future use cases might include linking existing schemas to the core ontology and developing domain specific ontologies.
My second question is: *Can we separate the definition of terms from the ontologies? * An example might make this clearer. We may be able to define TaxonName, GenusName and SpecificEpithet as terms and get general agreement on their meaning. This is separate from defining an ontology that stipulates TaxonName can only have a SpecificEpithet if it has a GenusName. One community may choose to use an ontology (or schema) to enforce this behaviour another might not but it doesn't change the meaning of the terms. DublinCore is like this. It defines a set of terms (as RDF properties in one form) that can be bolted together any way a user chooses but then the PRISM standard enforces their use in a certain way. Could decoupling of core terms from ontologies be a way forward?
I am willing to stick my neck out and put the first list of objects/terms together but we then need to discuss use cases and modify any initial list. I believe this will then feed into protocols and serializations etc..
Once again I am grateful for your thoughts,
Roger
Roger,
So my first question is: How do we develop a core vocabulary/ontology in a distributed, collaborative way?
In the absence of a more sophisticated tool (free, open source, capable of exporting models in different useful formats, allowing collaborative work, etc.) a wiki should be fine.
Regarding documentation, a mix between the CIDOC and DublinCore approaches sounds good.
My second question is: Can we separate the definition of terms from the ontologies?
If I understood correctly, the question is "which level of detail should we address?". I think we should start as simple as possible, maybe not trying to address constraints that look too specific, but also not leaving everything "independent" (links are important). We could define the main classes and properties from different parts of the ontology. Then we could try to represent them in some format, create instances, experiment alternatives of integration and extension, etc. -- Renato
participants (2)
-
Renato De Giovanni
-
Roger Hyam