Consider two RDF graphs:
Now link these two graphs together, with A linking to A',
B linking to B', C linking to C', and D linking to D'. That is, let them be the same URI.
Now consider the case where all the links are preserved, but I remove all parts of the connected RDF graphs except for those that have one degree of separation from the connected nodes. From this I can see relations between the graphs to one degree of separation. This may make more sense later with an application.
In addition, if I only want the parts of the RDF graphs that are connected to each other, I can also do that.
Now consider the case where I let one RDF graph be some reflection of what connections I see between what I know (Brent's Domain), and another graph be the connections between things relating to a project (Project Domain).
Does this seem useful? Now if I draw connections between URIs that are common to both Brent's Domain and the Project Domain I can see the things I know that apply to the project. Moreover, if I allow a few degrees of separation I can relate what I know to any URI describing the project. In this way, I may be able to come up with a plan of what I need to learn to understand a particular part of the project.
I could take this idea further. What if I replaced the Project Domain graph with an RDF graph describing another person?
This really is nothing new. Liyang Yu describes A Smart Data Integration Agent in the first chapter of his book, A Developer's Guide to the Semantic Web. His description mirrors the presented idea in form. Moreover, Liyang Yu describes the linking of URIs as distributed information aggregation.
There is one issue that must be considered of course. The URIs we choose must be describing the same thing. Perhaps something like regular expressions are in order would help people do this. Could the paper, "Processing SPARQL queries with regular expressions in RDF databases" by Lee et. al be useful?