The first technique that comes to mind is the URI. But even a URI isn't a complete enough description of how to get an even many web resources. Consider:
And that's just URI based locations. Obviously there will be custom descriptions for geographics and other physical locations.
I'm thinking it might be good to not refer to ideas as non-rdf concept (hash) URIs, even though URIs represent ideas.
Suppose I post an image at the URI <http://example.org/photo1>. I want to attribute a creator to the photo that the collection of pixels returned by dereferencing that resource represents. If the URI represents a concept, which one?
I'm guessing the last one is closest to our intention. This could have a traditional URI that would return a representation, (a PNG for example), but how would we then decide if we described the author of the art, the PNG, or the URI (location)?
If we want to describe an digital camera image, does it need its own idea URI (<http://example.org/photos#1>)? I'm thinking so.
Maybe not. Of course if the thing we actually mean to talk about is a URI, then the URI will do just fine. But again, what do we mean when we talk about the URI? I tend to think of it as location directions.
Every traditional resource we'd like to talk about has three parts
When do we know that a concept is being overloaded, that another related concept should be created?
I'd like to develop a pragmatic approach, where concepts bud into existance only when they have to, based on something algorithmic so you don't have to think about ontology. Something more biological and less philosophical.