A new algorithm for removing node overlapping in graph visualization Xiaodi Huang a, * , Wei Lai b , A.S.M. Sajeev a , Junbin Gao c a Department of Mathematics, Statistics and Computer Science, The University of New England, Australia b School of Information Technology, Swinburne University of Technology, Australia c School of Information Technology, Charles Sturt University, Australia Received 10 January 2006; received in revised form 8 February 2007; accepted 13 February 2007 Abstract Techniques for drawing graphs have proven successful in producing good layouts of undirected graphs. When nodes must be labeled however, the problem of overlapping nodes arises, particularly in dynamic graph visualization. Providing a formal description of this problem, this paper presents a new approach called the Force-Transfer algorithm that removes node overlaps. Compared to other methods, our algorithm is usually able to achieve a compact adjusted layout within a reasonable running time. Ó 2007 Elsevier Inc. All rights reserved. Keywords: Graph layout; Overlapping nodes; Force transfer; Layout adjustment 1. Introduction In the field of graph visualization, nodes in a graph represent objects or entities, which often have distinct labels as their identifiers. These labels in a drawing may be in the form of text, digits, or even images. Unlike most existing drawing algorithms, nodes should be drawn as rectangles that have enough area to display labels, rather than as abstract points with almost no size. UML diagrams in CASE tools, for example, are labeled graphs [5]. The problem of node-overlapping may arise while displaying such graphs using traditional algorithms [1]. The resulting layout contains overlapping nodes, which destroy the layout aesthetics – an underlying purpose of graph layout. This is because most of these algorithms do not take into account the node size. The need thus arises for removing node-overlapping. Furthermore, in a dynamic situation where a graph is often changed, such as enlargement/shrinkage of sub-graphs and addition/deletion of nodes, the layout should be adjusted accordingly. More importantly, such an adjustment for eliminating the clutter of a layout should be kept to a minimum. The techniques for removing node-overlapping can be applied directly to reposition overlapping windows in multi-window applications, to avoid the overlaps of compound nodes in 0020-0255/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.ins.2007.02.016 * Corresponding author. Tel.: +61 2 6773 2118; fax: +61 2 6773 3312. E-mail address: xhuang@turing.une.edu.au (X. Huang). Information Sciences 177 (2007) 2821–2844 www.elsevier.com/locate/ins