Interactive knowledge discovery and knowledge visualization for decision support in multi-objective optimization

In many practical applications, the end-goal of multi-objective optimization is to select an implementable solution that close Pareto-optimal front while satisfying decision maker’s preferences. The making process challenging since it involves manual consideration all solutions. field multi-criteria offers methods help maker in this process. However, most only focus on analyzing solutions’ objective values. A more informed generally requires additional knowledge how different preferences affect variable One difficulty realizing are often expressed space, required implement a preferred exists space. paper, we propose support system allows interactive discovery and visualization practitioners by simultaneously considering space their impact step can use either two recently proposed data mining techniques for extracting rules conform given preferences, extracted visualized via novel graph-based approach important variables, values interactions with other variables. result intuitive aids entire — from visualization. We demonstrate usefulness benchmark problems up 10 objectives real-world six objectives.