The self-organizing map's unsupervised clustering property, is known for classifying high dimensional data sets into clusters that have similar features. Using this property and arranging self-organizing maps into hierarchies, we demonstrate in this paper that legacy code can be potentially broken down into suggested classes using hierarchical self-organizing maps. This is in conjunction with inheritance that is typical in the object oriented approach. The results shown from our study indicate common features between procedures at di#erent levels of the self-organizing map hierarchy structure. This suggests that the self-organizing map is capable of proposing the di#erent levels of classes that can be extracted from legacy code, aiding the process of software reverse engineering.

Analysing high-dimensional data is a task where software tools can reasonably assist the data analyst, by visualising, and thereby uncovering, the inherent structure and topology of the data collection. Especially the kinds of tools that can produce results autonomously, i.e. unsupervised tools, are

The self-organizing map’s unsupervised clustering method can be used as a data visualisation technique. Within this paper different techniques to visualise selforganizing maps (SOM) and their effectiveness are investigated in relation to the organisation of a large software collection and its visualisation. GENISOM, an offspring component of the GENESIS software engineering platform, incorporates the generation, maintenance and viewing of Self-Organizing Maps. The results from our studies indicate that a hybrid of 2D and 3D visualisations is favoured by users. Extensive usability tests also show that the majority of users found that the additional information a SOM provides, aids browsing and searching of a software collection. Further work is addressing the problems found in the application of SOM within a software engineering environment.