Dynamic Graph Drawing of Sequences of Orthogonal and Hierarchical Graphs CarstenG¨org 1 , Peter Birke 1 , Mathias Pohl 1 and Stephan Diehl 2 1 Saarland University, FR Informatik, D-66041 Saarbr¨ ucken, Germany 2 Catholic University Eichst¨att, Informatik, D-85072 Eichst¨att, Germany goerg@cs.uni-sb.de, diehl@acm.org Abstract. In this paper we introduce two novel algorithms for draw- ing sequences of orthogonal and hierarchical graphs while preserving the mental map. Both algorithms can be parameterized to trade layout qual- ity for dynamic stability. In particular, we had to develop new metrics which work upon the intermediate results of layout phases. We discuss some properties of the resulting animations by means of examples. 1 Introduction In many applications graphs are not drawn once and for all, but change over time. In some cases all changes are even known beforehand, e.g. if we want to visualize the evolution of a social network based on an email archive, or as in this paper the evolution of program structures stored in software archives. In these kinds of applications each graph can be drawn being fully aware of what graphs will follow. Unfortunately, to the best of our knowledge there exist only two algorithms that take advantage of this knowledge, namely TGRIP [6] and Foresighted Layout [8]. See Section 6 for a discussion of these and other approaches. While the former was restricted to spring embedding, the latter is actually a generic algorithm. Recently we introduced Foresighted Layout with Tolerance (FLT) [7] for drawing sequences of graphs while preserving the mental map and trading layout quality for dynamic stability (tolerance). The algorithm is generic in the sense that it works with different static layout algorithms with related metrics and adjustment strategies. As an example we looked at force-directed layout. In this paper we apply FLT to orthogonal and hierarchical layout, which means that we have to develop adjustment strategies and metrics for these. We also improve FLT by introducing the importance-based backbone as a generalization of the supergraph of a sequence of graphs. 2 Improved adjusted Foresighted Layout In our previous work the supergraph, which is the union of all graphs in a graph sequence played a crucial role. The reason for using the supergraph was that it provided all information about the graph sequence and that its layout could Proceedings of 12th International Symposium on Graph Drawing, New York City, USA, September 29 - October 2, 2004