7 February 2013
Applications of network analysis to large-scale software systems
12:30pm - 13:45pm
Dr Peter Andras - Newcastle University
Large-scale software systems include many components, possibly written by different teams and different times, making their maintenance difficult. Their sheer size makes difficult the application of many traditional software engineering techniques in the context of maintenance of such systems. A further complication is that the dynamic behaviour of such systems may be difficult to predict on the basis of static analysis of their code. Complex systems analysis methods, like network analysis, originating from the investigation of biological and social systems, offer a promising approach to gain a better understanding of such large-scale software systems. We applied network analysis methods to dynamic analysis data derived from large scale software systems (e.g. Google Chrome, JHotDraw, aTunes, Mars – Sim, etc.). We investigated the identification of functionally important methods that are relevant for given usage scenarios or software features of the system. We also analysed the relationship between method prototypes derived from static analysis and the local dynamic network characteristics of the methods. The results are promising and show that network analysis methods can help the understanding and maintenance of large scale software systems.
Peter Andras did his undergraduate and postgraduate studies at the Babes-Bolyai University, in Cluj, Romania. He moved to Newcastle University in 2000, first to the Department of Psychology and then in 2002 to the School of Computing Science. His background is in machine learning, computational neuroscience and complex systems. Currently he is a Reader in complex systems and his research group works on applications of complex systems analysis to large scale software systems, biological neural systems, and large-scale bio-medical data. He is Senior Member of the IEEE and Fellow of the Society of Biology.
Save to your Calendar