VisAware (PDF link) is the generalization of a visualization technique for situational awareness, developed by Yarden Livnat, Jim Agutter, Shaun Moon and Stefano Foresti, all of the University of Utah. I'll attempt a bad capsule summary. A spatial map is ringed with concentric circles. The circles capture time and type of event occurrences. Lines connect urgent events with locations on the map. Connect with color and interactivity and you get an effective display for keeping on top of a large number of events without an a priori obvious way to correlate them, but three potential correlation vectors: time, type and location. While the screen captures are exceedingly seductive, the paper describes the rigorous construction of a formal framework for generating them.
The authors aren't particularly clear on what this technique isn't good for, but an extrapolation to webfeed aggregation is probably worth a prototype. Take a relatively large blogroll and use some clustering techniques to generate a two dimensional map between them. Then blog posts become the events, which conveniently have a time element, and can point back to feeds to which they're related. Event typing could be hardwired to some levels of "importance", or user controlled, e.g. the result of standing searches against aggregated items.