A topology-shape-metrics approach for the automatic layout of UML class diagrams

Abstract

Class diagrams are among the most popular visualizations for object oriented software systems and have a broad range of applications. In many settings it is desirable that the placement of the diagram elements is determined automatically, especially when the diagrams are generated automatically which is usually the case in reverse engineering. For this reason the automatic layout of class diagram gained importance in the last years. Current approaches for the automatic layout of class diagrams are based on the hierarchic graph drawing paradigm. These algorithms produce good results for class diagrams with large and deep structural information, i.e., diagrams with a large and deep inheritance hierarchy. However, they do not perform satisfactorily in absence of this information. We propose in this work a new algorithm for automatic layout of class diagram which is based on the topology-shape-metrics approach. The algorithm is an adaption of sophisticated graph drawing algorithms which have proven their effectiveness in many applications. The algorithm works as well for class diagrams with rich structural information as for class diagrams with few or no structural information. It improves therefore the existing algorithms significantly. An implementation of the algorithm is used in the reverse engineering tool JarInspector.

Assessing the Comprehension of UML Class Diagrams via Eye Tracking

“…Tilley et al [Tilley, Huang 2003] investigated the use of UML syntax, semantics, spatial layout, and domain knowledge in system evolution tasks. Eiglsperger et al [Eiglsperger, Kaufmann, Siebenhaller 2003] proposed an automatic layout algorithm for UML class diagrams. Their algorithm is based on the topology and shape metrics that try to minimize the edge crossing, bends, and occupied area.…”

“…Definition:The orthogonal layout focuses on the minimization of the edge crosses and bends. Multiples of 90 degree angles are used to position the intersecting edges [Eichelberger 2003;Eiglsperger, Kaufmann, Siebenhaller 2003;Purchase, Allder, Carrington 2002]. This layout is adopted from general graph drawing algorithms and is typically available in UML modeling and drawing tools.…”

“…Representations, including layouts of UML class diagrams, are a general research topic with regards to software comprehension and maintenance activities. Investigations in the software visualization and program comprehension communities have primarily focused on effective layout schemes [Andriyevska et al 2005;Eiglsperger, Kaufmann, Siebenhaller 2003;Sun, Wong 2005] and key aesthetics criteria [Eichelberger 2002;Eichelberger 2003;Gutwenger et al 2003] with the goal of aiding and/or enhancing the human cognitive process. A number of usability studies have been reported in the software visualization research literature that evaluate UML class diagrams, including those with additional semantic information (e.g., class stereotypes), for an effective representation in addressing various software evolution tasks [Andriyevska et al 2005;Arisholm et al 2006;Kuzniarz, Staron, Wohlin 2004;Staron, Kuzniarz, Thurn 2005].…”

Empirical assessment of UML class diagram layouts based on architectural importance

“…An aesthetically pleasing class diagram might look pretty but it increases the cognitive workload of the person reading it. A variety of UML class diagram tools [Eichelberger 2001;Eiglsperger, Kaufmann, Siebenhaller 2003;Gutwenger et al 2003b] have been developed but relatively little effort is put into evaluating the effectiveness and utility of the tools. This work bridges the gap in empirical evaluation of UML class diagram layout techniques.…”

CVSscan

“…The preventive approach tries to improve the reliability of a system at design time. Many visual tools and techniques exist to improve the expressiveness of UML and visually assess design-time quality attributes [Eiglsperger et al 2003;Gutwenger et al 2003]. The corrective approach aims to facilitate the maintenance phase, and is supported by program and process understanding and fault localization tools (e.g.…”