A Low-Effort Animated Data Structure Visualization Creation System

Abstract

Algorithm animations and high level data structure visualizations are not widely used, in part because producing them is difficult and time consuming. Algorithm animation and data structure visualization systems attempt to minimize this effort, but none has made the process as simple as it could be. The purpose of the system described in this work, the jGRASP Visualization System, is to allow the creation and use of dynamic data structure visualizations from working source code with minimal effort. Ideally, the only work required to create an animated visualization should be selecting values from a program running in debug mode in an IDE, selecting the way each value will be displayed from a list of available choices, and physically arranging the display elements. This has been achieved for arbitrary implementations of common data structures in Java. This dissertation begins by discussing existing work on the usefulness of algorithm animation and data structure visualization for algorithm and code understanding, and existing work on systems with similar organization, features, and construction to the one described here. The jGRASP Visualization System is described in detail from the user's perspective. Implementation details are discussed. Previously published results on the effectiveness of these visualizations when applied to arbitrary data structure code, and on code understanding experiments performed with the visualizations are presented. Finally, visualization feature usage data collected from users of the system is analyzed and discussed.