Vlado, thanks for you feedback. Before I add your comments let me ask a couple clarifying questions to make sure I understand your additions.
What's new on this subject? veiwpoint? -I understand why this is useful, but is this really related to social context. Social context of content does not require categorizing the subject of viewpoint of content.
node A -Fosdem node B -Fosdem node A and node B might not have similar subjects. Do you mean what's new on this social context? i.e if node C -Fosdem then you could see the latest content for a social context of Fosdem?
Decentralised content discovery? Correct. This is the major thrust, can we apply a common set of labels across data sources with common dimensions(descriptions) so that we can discover social context.
Example: Users table Contact table Contact manager table Buddylist table XFN hypthetable CiviCRM-table Profile-table They could all have an e-mail field and be mapped to a common label e- mail, with a common description. Therefore any new data source say Aura, that complies with the e-mail label, and description could be discovered through a cross query of that data source. However, that capability would fall clearly outside what was possible for a summer's worth of development.
Content propagation? I suspect that content propagation is outside the scope of social context but please explain further if you think it is relevant. Discovery of a context neighbourhood? I assume you are talking about some form of statistical technique for weighting or clustering common social context terms. Great idea, but I think that just getting the social content to be manually added and extracted from solid relationships will be a summers worth of work.
Several of these are great Phase III, Phase IV type projects. I'd be happy to add them if you wanted to help flush out the deliverables, that would probably have to go to other students. Cheers, Kieran