Lambda the Ultimate

Hi,

I have an idea for an academic paper, and am seeking help for prior art and research.

My idea is about proof witnesses, and what to do when a group of witnesses contradict each other (software bug/logical inconsistency), and repairing the proof. Without going into details on the algorithm, the metaphor is simply to split witnesses up, using a mathematical plane to draw a group of witnesses into two groups: definitely lying (the witness whose account is proven contradictory/false), and somewhat certain (witnesses with similar testimonies who have not yet been proven contradictory/false, but might once asked further questions or may be proven "good enough" given the new questions/evidence we've asked the witness about). My idea comes from witness interrogation techniques in the real world.

Since I haven't been active in PLT for several years, my knowledge of what to search for and keywords to use on ACM and IEEE search engines is pretty weak, and I could use some tips.

Thank you very much!