When drawing graphs whose nodes contain text or graphics, the nontrivial node sizes must be taken into account, either as part of the initial layout or as a post-processing step. The core problem in avoiding or removing overlaps is to retain the structural information inherent in a layout while minimizing the additional area required. This paper presents a new node overlap removal algorithm that does well at retaining a graph's shape while using little additional area and time. As part of the

more »

... alysis, we consider and evaluate two measures of dissimilarity for two layouts of the same graph.