How would you extract actionable knowledge from sub-module test results and scale it to full-system conclusions?
