How to get exactly the definition I need and nothing more..Is MIREOT the answer?

I am back to writing small scale ontologies for my research prototypes. In the interests of reusability, I would like to extend definitions of my atomic concepts from concepts in high level ontologies such as the Basic Formal Ontology (BFO), the Information Artifact Ontology (IAO), and the General Formal Ontology (GFO). I would like to reuse relation definitions from the Relation Ontology from the OBO Foundry and Michel Dumontier's extensions. However, a simple import on my ontology development tool (TopBraid) for any of these ontologies is enough to bring the whole application to its knees. Let us say I want to define a concept Fusion as a subconcept of the high level ontology concept Chronoid from GFO. Fusion is the root concept of my ontology and I only need to extend one concept from GFO. Instead, I am stuck with importing the ENTIRE GFO ontology, lock, stock, and barrel. With no other alternative, I am stuck with creating "concept silos," isolated from the rest of the Semantic Web. Rather ironical. Wouldn't it be nice to use just what I want and leave everything else out?

That was the idea behind MIREOT (Minimum Information to Reference an External Ontology Term), a nice little workaround developed by the good folks at the OBI (Ontology for Biomedical Investigations) workgroup. MIREOT can actually used as a verb by as in, "I MIREOTed the BiologicalProcess concept from the Gene Ontology." Compare that with importing all of GO! MIREOT has been around for a few years now, however, there seems to be no available MIREOT plugin for popular ontology development tools like Protege and TopBraid. I'm not sure about the Sentient Knowledge Explorer from IO Informatics. This was a very nice looking tool I was introduced to during my post doctoral stint with Mark Wilkinson at UBC. This may be something worth looking into.

Another idea. Why do we need to import these huge ontologies in the first place? Given network connections speeds today, wouldn't be even easier for an agent to check a URI to confirm a referenced concept exists remotely, and create a subconcept or even a subrelation link from the referencing ontology to the referenced concept? This would work just same as a hyperlink from Web 1.0. Thoughts?