Edge bundling is a technique commonly used to reduce visual clutter and improve the comprehension of the drawings of large graphs. Here, we model edge bundling as a multi-objective optimization problem and employ clustering strategies, metaheuristic and Pareto analysis to identify non-dominated solutions for some classical graphs from the literature.

One of the commonly used techniques to improve the readability of large graph drawings is called edge bundling, which groups edges in such a way that reduces the visual complexity of the drawing. This paper proposes to treat this task as a clustering problem, using compatibility metrics to evaluate the generated solutions in an optimization pipeline, combined with a clustering ensemble approach. The goal was to solve the General-based Edge Bundling (GBEB) problem with relatively low computational costs using a method called Clustering Ensemble-based Edge Bundling (CEBEB) and evaluate the results. CEBEB proved to be a very promising alternative to solve GBEB, since it is capable of generating relatively good solutions with shorter run-times compared to an existing, well-established GBEB method.