A modular attachment mechanism for software network evolution

Exploring community structure of software Call Graph and its applications in class cohesion measurement

“…These researches have revealed that networks that are constructed from software systems usually have Scale-free degree distributions (Myers, 2003;Potanin et al, 2005;Baxter et al, 2006;Louridas et al, 2008), and exhibit Small-world properties (Myers, 2003;Valverde and Solé, 2007;Qu et al, 2015), like other typical complex networks. Researchers have also found that the evolution processes of these networks are in accordance with the predictions made by Complex Network theory (Pan et al, 2011;Turnu et al, 2011;Li et al, 2013). These results have shown that software systems usually exhibit typical characteristics of complex network systems, thus have laid a good foundation for us to further use community detection algorithms that originally developed in Complex Network theory to analyze software systems.…”

“…In recent years, the theory has been successfully applied in the domain of software engineering, including software evolution process modeling and understanding (Pan et al, 2011;Turnu et al, 2011;Li et al, 2013), software evolution prediction (Bhattacharya et al, 2012), software structure interpretation and evaluation (Myers, 2003;Potanin et al, 2005;Baxter et al, 2006;Louridas et al, 2008), software execution process and behavior modeling (Cai and Yin, 2009;Qu et al, 2015), etc. These researches have revealed that networks that are constructed from software systems usually have Scale-free degree distributions (Myers, 2003;Potanin et al, 2005;Baxter et al, 2006;Louridas et al, 2008), and exhibit Small-world properties (Myers, 2003;Valverde and Solé, 2007;Qu et al, 2015), like other typical complex networks.…”

Forecasting Java Software Evolution Trends Employing Network Models

“…Li et al [49] proposed a model according to which, software networks grow not only by individual node additions, but also with multiple node attachments in the form of complete modules. The model starts with the insertion of single nodes in the existing network, but after some initial steps the number of nodes added in each step is increasing.…”

“…Furthermore we model the creation of edges between existing nodes, between existing and new nodes as well as the removal of existing edges, issues that have not been taken into account in previous models but which fundamentally affect the topology of the resulting software networks. Finally, the evaluation is performed on ten open-source systems (models proposed by Zheng et al [93] and Li et al [49] are evaluated on a single system), which enables the generalization, up to a certain degree, of the observed evolutionary trends in software evolution.…”

“…Modeling software systems as networks enabled a graph-based treatment and analysis with the goal of investigating several properties, such as scale-freeness [53], [59], [67], [73], [85] and the presence of small-world phenomena [40], [77]. However, with the exception of few recent research efforts [8], [49], [93], most of the studies that employ graphs for the representation of software, focus on static snapshots of the examined systems in the sense that individual software versions have been analyzed without paying attention to the evolution of the corresponding networks.…”