Visualization of metrics and areas of interest on software architecture diagrams

Promotie: dhr. H.V. Byelas, 13.15 uur, Academiegebouw, Broerstraat 5, Groningen
Proefschrift: Visualization of metrics and areas of interest on software architecture diagrams
Promotor(s): prof.dr. A.C. Telea
Faculteit: Wiskunde en Natuurwetenschappen
Contact: via Klazien Offens, tel. 050-363 8079, e-mail: k.m.offens@rug.nl

Visualization of metrics and areas of interest on software architecture diagrams

Software structure is depicted with architecture and design diagrams. One diagram typically emphasizes on the particular aspects, functionalities, and collaborations of selected system parts. We call such parts areas of interest (AOIs). One diagram can contain many overlapping AOIs, and one AOI can occur in several diagrams. Hence, AOIs cannot be effectively shown using standard structural diagram notations.

Software quality is typically captured using software metrics computed on diagram elements (functions, classes, packages, and groups thereof).

Effective system understanding requires ways to correlate several metrics, defined on several elements, with the system structure and AOIs. In this thesis, we introduce new ways for effectively visualizing AOIs and their associated metrics. We use a combination of geometric annotations, shading, and texturing that attempts to mimic how humans draw AOIs on paper diagrams during design sessions. Secondly, we extend table visualization techniques to display multiple metrics defined on individual diagram elements. Overall, we combine structural and metric information defined at several levels of detail in a single visual representation, thereby addressing assessment tasks that require correlating multiple data sources. Although our interest is in software architecture diagrams, our methods can be used in other contexts, such as organization charts or spatial maps. Throughout this work, we conducted evaluation studies of the proposed visualization methods, both in academic and industrial contexts. We discuss how the results hereof have been used to improve our visualization design and validate our design choices.