Abstract: |
This document describes a preliminary PhD thesis proposal that will hopefully lead to a collaborative framework and platform for Software Evolution Visualization (SEV).
We sustain our decision to follow this research area by evaluating recent works which have shown that there is a need for multi-metrics, multi-perspective and multi-strategy approaches to SEV as summarized by (Novais, et al., 2013). The authors identify some research niches such as missing case studies, tool comparisons and experiments with the aim of predicting defects, improve software quality and development processes. Another missing aspect relates to the presentation of real scalable visualization and dependency impact among projects. It is also recognized that there is little formal validation and collaboration in this area, most likely because the data is scarce, dispersed and not widely shared by each individual researcher. The lack of empirical studies is a real constraint to allow the community to perform benchmarking and compare methodologies and results. In other words, the SEV community has failed to provide sound evidences, through empirical validation studies, of the impact of using the technology they created. In fact SEV research deliverables provide visual insights that are expected to help understand complex software artefacts and ultimately contribute to improve their quality and the maintenance process itself.
The goals of our research consist on proposing a structured approach to (i) collect data from public domain software repositories, (ii) extract complexity and quality metrics using a meta-model driven measurement approach (M2DM), (iii) store and eventually transform those metrics by adopting big data technologies for scalability sake, (iv) visualize software evolution, along the corresponding metrics, in a collaborative fashion, allowing to identify patterns and trends. The aforementioned approach is expected to scaffold exploratory activities on top of the collected data, allowing the community to do benchmarking, evaluate software engineering best practices and assess software engineering research questions by means of empirical studies (Goulão, et al., 2012).
|