Closed bplewe closed 8 years ago
Sure. Let's do it. Why don't you find some space on the wiki and let's see what you've got.
Bump.
I'm ready to start some actual work on this. I'm wondering how to best represent the graph. It seems like RDF/XML is still the prominent representation format for vocabularies, but schema.org uses RDFa.
Any feedback/input?
Ryan, sorry, I backed off a bit from this; it is hard from one day to the next to tell whether semantic web/RDF/linked data is the wave of the future or practically dead :). RDF/XML is almost universally hated, and RDF 1.1 officially supports several serializations; personally, I think TRiG is the most promising (named graphs/quads may be necessary to make this work). RDFa is only useful for attaching attributes to HTML pages. Personally, what I am after is a way to build a linked data front end on FamilySearch Family Tree. Heck, maybe JSON-LD is the way to go?
Here is the missing link
http://manu.sporny.org/2014/json-ld-origins-2/
and a thumbs up for JSON-LD.
Joel
I finally got around to publishing some semantic schema definitions for the GEDCOM X conceptual model. These definition documents are really just a different encoding of the normative definitions provided by the GEDCOM X Conceptual Model specification. The Conceptual Model Specification officially provides the canonical definitions, but these semantic schema documents will make it easier for structure data (e.g. RDFa, Microdata, JSON-LD) processors to use.
I chose to use RDFa as the encoding for these schema definitions, following the lead of schema.org. I can certainly imagine providing other encodings of the definitions (e.g. Turtle, JSON-LD, RDF/XML), but we'll start with RDFa for now.
One of the nice benefits of this effort is that you don't get a 404 now when you make HTTP requests to the various GEDCOM X data types and controlled vocabulary elements. Instead, the request resolves to one of these semantic schema definitions. For example:
This may be a sore subject, but I'm wondering if the spec is now mature enough to redevelop RDF as a serialization option. I'm new, but I heard it was in the conceptual model before, and that didn't go well.
I have a genealogy project that demands a linked data solution, so if the community doesn't develop a genealogy ontology, I'm going to have to. I don't have 10 years to wait for FHISO to get something done, so GEDCOM X seems like the only game in town. In my experiments so far, it should be doable with the conceptual model as-is (setting up equivalencies to FOAF, BIO, DC, etc.). My early experimental Turtle files have been 30-40% shorter than the XML and JSON files.
Has anyone been working on this? Anyone interested? Can I post the files to this GitHub?