MDS-Views : Visualizing Problem Report Data of Large Scale Software using Multidimensional Scaling ∗

Gaining higher level evolutionary information about large software systems is key a in validating past and adjusting future development processes. In this paper we address the visualization of problem reports by taking advantage of the proximity introduced by changes required to fix a problem. Our analyses are based on modification and problem report data representing the system’s history. For computation of proximity data we applied a standard technique called multidimensional scaling (MDS). Visualization of feature evolution is enabled through the exploitation of proximity among problem reports. Two different views support the assessment of a system design based on historical data. Our approach uncovers hidden dependencies between features and presents them in easy-to-evaluate visual form. Regions of interest can be selected interactively enabling the assessment of feature evolution on an arbitrary level of detail. A visualization of interwoven features can indicate locations of design erosion in the architectural evolution of a software system.