Interactive 3D visualisation of software metrics

Abstract

Nowadays, companies exist in a very dynamic and highly competitive market, which leads to ever-growing software requirements that change continuously, even over a relatively short span of time. Consequently, software is becoming more and more complex, giving rise to issues of long-term stability and maintainability. The key solution is to approach software in a more rigorous manner. Thanks to the work of people like Chris Kemerer and Shyam Chidamber [14], one can now empirically analyse this realm with the help of software metrics. Software metrics provide visibility into the characteristics of software systems and help expose certain issues in software, which lead to better risk analysis. However, when dealing with large-scale systems, having a large number of metric results may lead to cognitive overload and make problematic areas difficult to find. This project explores the use of 3D visualisation as a means of reducing cognitive load and facilitating the interpretation of software measurement data for non-trivial systems. This involves the development, and evaluation, of a system that implements and evaluates a 3D interactive visualization centred around a city metaphor [24]. The project extracts object-oriented metrics from a software system and divulges their semantic meaning through entities in the virtual cities, representing the different classes within the software project as buildings, and their respective metric counts as attributes of these buildings, such as colour and height. This way, the human mind automatically processes the metrics intuitively and efficiently. The concept and implemented artifact are criticised by contrasting visualisations of several versions of the same system. The merits of the results achieved are discussed based on the visualisation metaphor's effectiveness in helping the user understand metric values over successive versions of a system with minimal cognitive effort.