I'm writing a Log-Event correlator. The principle is to have several log-sources analyzing the same signal in multiple places, and then correlate all events to determine causes by method of least-common-denominator.
The output becomes something like; (event-count can be expanded to actual events for drill-down) nodeA,nodeB,nodeC: 524 nodeA,nodeB: 65 nodeA: 15 nodeC: 64
The events is likely to partition the nodes such that you won't find correlated events both in A,B, and B,C, but it could happen under rare circumstances.
I would like to visualize this in some suitable way. I'm considering some kind of "venn"-diagram on stereoids, but I wanted to hear if someone have other ideas, or know of similar work anywhere else?