Planning an Experiment on User Performance for Exploration of Diagrams Displayed in 21⁄2 Dimensions

Abstract

Two dimensional node-link diagrams such as those proposed by the Unified Modeling Language (UML) have become an important means of communication and documentation of various aspects of software systems. When complex systems are concerned, these diagrams typically consist of a great amount of connected nodes – particularly if they origin from reverse engineering processes. In software visualization research community various papers have been published arguing that displaying diagrams in 3D or 21⁄2D, i.e., on a plane seen under 3D perspective, helps to cope with the size of the diagrams and supports users in exploring them. However, formal experiments supporting the hypothesis that user performance increases when analyzing diagrams in 3D or 21⁄2D instead of 2D is usually missing. In this paper we describe a formal experiment we are about to conduct to examine user performance for tool-supported exploration of control flow diagrams displayed in 21⁄2D.