Oh I see, basically I have to create my own graph describing where <some-graph-uri> originated from. This makes perfect sense. There is however another problem that comes to mind, although this is strictly theoretical because I don't have this problem in practice (yet). Let's say I want to import the datasets LOV-1 and LOV-2 (I just chose two random names). For example they could be 2 graphs maintained by 2 communities. What if both datasets define the same label? Or if they have duplicate (s, p, o, g) quads? How do I distinguish the URI in LOV-1 from the URI in LOV-2?
==================================================== The way to approach this is to think of "tagging" the graphs rather than the quads. For each of your graphs you have a URI to identify it. This means you can (in another metadata graph) now made RDF assertions about that URI. So you can attach as many category, tag and label values as you like that way. You can use queries to find the graph URI matching your tag set of interest and then retrieve the target information from that graph. Dave