In preparation for the Nomina session on Thursday.
Here are some things to look at for the session titled "Exploring Semantic Web Technologies for Global Names Architecture Name Resolution"
Dima and I have figured out a way to expose the name strings in the GNI to the Semantic Web.
At this point it is really about exploring ways in which distributed data projects can link to entities in the GNI.
In these experiments we have been exploring the mapping of names strings to species concepts.
For the most part these should work with different kinds of species concepts but for now this experiment is using the TaxonConcept.org concepts.
The initial stages were fairly basic. This concept has some association with the following name strings in the GNI.
There was no real effort to determine the nature of these relationships until very recently.
I marked up an RDF file that makes the following relationship assertions about how a concept is related to a particular name string.
The GBIF site mentioned a movement toward databasing names in the following plain text form Puma concolor (Linnaeus 1771)
* Italics can be added in web views etc.
For the sake of argument, lets say that the preferred from is like that above with no comma between the author and date.
This particular string could them be mapped to slight variations that are correctly formed and mean the same thing.
This particular string could also be mapped to common misspellings etc of the same name.
Those various alternatives form a lexical group.
Similarly, the synonym Felis concolor Linnaeus 1771 also has a list of essentially correct strings of characters and common malformed variants.
This would be a second lexical group.
So the concept represented by this page and URI can then be related to these name variants and a linkage is formed between the
entities in the TaxonConcept Dataset and the GNI dataset.
These relationships can be represented in the N3 form which some humans find easier to interpret than the alternative RDF/XML form.
Note that these predicates have the "URI" suffix to differentiate between those predicates that expect URI object and and similar predicates that expect a literal object.
Subject Predicate Object
Rather than rely on inferencing I also make the reverse relationship
When exposed to the Linked Open Data Cloud those statements are associated with both the species concept and the GNI namestrings.
In my own quad store you can see that they appear in this view of the species concept:
Not that relevant statements from other databases also appear - demonstrating one of the key advantages of the Linked Data approach.
The nomina RDF and other information is part of this view even though it was not part of the basic species concept RDF above.
The same concept is viewable on the cloud via the prettier Sig.ma service.
Note Sig.ma service also resolves the GNI namestring URI and exposed the rdf:label associated with that URI.
- Pete