Each collaboration between two developers (A and B) in the software development network is considered as a possible source of architectural and design decision, therefore a developer is considered aware of a decision if he or she is strongly connected to at least one of the two developers whom generated the decision. In-communicability is related to every collaboration within the collaboration DSN and it is based on Tamburri et al.’s formulation:
MAI = DEM - DAM
DEM = n_collaborators_of_two_developers/n_developers
DAM = n_collaborators_of_two_developers_who_communicate_with_them/n_developers
Therefore, global in-communicability can be defined as the mean MAI over the entire collaboration network. Communicability is a global indicator which consists n the mean of all local communicability measures, calculated for every collaboration between two developers within the collaboration Developer Social Network in the range in analysis. Communicability was preferred to in-communicability in order to simplify measurement comprehension, because in-communicability tend to be characterised by measurements that tend to zero. Communicability is computed as:
(1 - mean(incommunicability))
community_metric_mean_communicability(mail.graph, code.graph)
| mail.graph | A uni-modal network G = (V,E), where V are developers (authors or contributors) and E is a collaboration between developers in the mailing list |
|---|---|
| code.graph | A uni-modal network G = (V,E), where V are developers (authors or contributors) and E is a collaboration between developers in the git log. |
D. A. Tamburri, F. Palomba and R. Kazman, "Exploring Community Smells in Open-Source: An Automated Approach" in IEEE Transactions on Software Engineering, vol. 47, no. 3, pp. 630-652, 1 March 2021, doi: 10.1109/TSE.2019.2901490.
Simone Magnoni (2016). An approach to measure Community Smells in software development communities. (Doctoral dissertation, Politecnico Milano).