Evolutionary and classical algorithms are most often applied to find one or more high-performing solutions for an industrial problem. Such an effort takes a considerable time to formulate the resulting optimization problem, customize an existing algorithm to make it efficient to solve the problem in a reasonable computational time, and evaluate the obtained solutions for their sensitivities to parameter changes. While the final high-performing solutions are worth the whole effort and the ensuing collaboration between academic and industrial partners, the effort can be exploited by executing an additional task of extracting knowledge about key features associated to the obtained solutions. In this talk, we shall propose a generalized framework for extracting useful knowledge from the optimization process and simultaneously utilize the obtained knowledge to improve the performance of the optimization algorithm itself. In recent studies on the proposed “innovization” method and its variants, it is argued that in routine optimization applications, the discovery of problem knowledge stays as valuable information about the he solution procedure of the problem in general, in addition to arriving at high-performing solutions themselves for the specific problem.