| The visualization of information structured as a graph or network has applications in a variety of areas including programming and software development, database design, VLSI design, web navigation, network administration, modelling protein interactions and molecular structures, modelling social interactions between people and organizations, and illustrating train timetables. The problem of producing clear and readable drawings of graphs has received a great deal of attention from researchers.; We present work on several problems related to graph drawing, ranging from techniques for drawing algorithms to tools for making existing drawing algorithms easier to use.; In addressing compaction of orthogonal representations, we present a characterization of a class of orthogonal representations for which orthogonal drawings with minimum area or minimum area and total edge length can be produced efficiently. We also present heuristics for the general case which result in improved performance over previous compaction heuristics. These techniques can be used to improve existing orthogonal drawing algorithms based on the topology-shape-metrics approach.; The problem of measuring the similarity of two drawings of the same or nearly the same graph has received little attention. The need for such a measure arises in applications in which the user is working with a graph which changes over time and must be periodically redrawn. Ideally, if the graph structure only changes in a small way, the new drawing will only be a little different from the previous one. Designing effective drawing algorithms for this scenario requires an understanding of what “a little different” means. We begin the task of finding a suitable similarity measure by defining and evaluating several potential measures.; Finally, we present tools to facilitate the use of graph drawing technology by experts and novice users alike, in order to bring graph drawing to a wider audience. The Graph Drawing Server makes graph drawing technology easily available over the Internet, while GeomNet extends the idea to computational geometry algorithms. Both the Graph Drawing Server and GeomNet provide interfaces which can be used interactively or called as a subroutine by a program. PILOT is a learning tool based on these systems. |
