On the role of aesthetics in genetic algorithms applied to graph drawing

Despite the role that aesthetics plays in information visualization, it is often downplayed or ignored in favor of functionality. However, by understanding how graphical representations are perceived it is also possible to improve them and create more comprehensible data visualizations. Meaningful relationships and data patterns can easily get lost among the representation of large and complex datasets. Various methods have been created to reduce visual clutter by either sorting nodes to minimize the number of intersecting edges, or by grouping edges into bundles with clear directions. In information visualization, perception principles have started being integrated into evolutionary computation in order to solve aesthetic problems, as they are capable of looking for solutions that may be found beyond local optima. In this paper we present a study on the importance of aesthetics and how evolutionary approaches can be used to influence visualization. This is supplemented with two case studies involving the design of genetic algorithms for reducing visual clutter through edge crossing minimization and edge bundling parameter optimization.