Research On Developers' Behaviors And Interests Based On Human Dynamics And Complex Networks

The last few decades have witnessed the rapid development and adoption of in-formation technology in a variety of industries.Software,one crucial constitute of in-formation technology,drives the innovation of society and significantly improves our work and daily life.Nowadays,our daily and work life depends heavily on various kinds of software,from mobile phone apps to office software.Software and software developers become the driving force of social innovation,and it is of great significance to study the behaviors of developers and the relations between developers and software.Modern software engineering methodology encourages software developers to be social and open-source minded.GitHub,an outstanding open-source and social collab-orative coding platform,was launched on April 10th,2008.It provides abundant social functionalities such as watching and forking for developers to interact with projects ef-ficiently.Meanwhile,GitHub has collected abundant data about online behaviors of developers and various kinds of interaction relations among developers and between developers and software projects.We study the behaviors of developers from the perspective of human dynamics and we also study the relations between developers and projects to reveal their interests from the perspective of complex networks.The empirical analysis on developers' blogging behavior based on human dynam-ics shows that developers' blogging actions follows a heavy-tailed distribution,indicat-ing that the majority of developers are less active yet a small number of developers are highly active.The result also shows that the heterogeneity of commit number distribu-tion of GitHub Pages is less than normal online systems such as movie rating,bookmark collecting,online auction,online message-sending,and twitter posting while the inter-event time distribution of developers' blogging behavior is more heterogeneous than Sina Weibo posting.We also find that the majority of developers are of low memory and high burstiness while elite developers are of low memory and low burstiness,indi-cating that the elites' interest on blogging is a long-term interest,and they even treat it as a regular work.We model the multiple relations between developers and projects as a multilayer bipartite network and study the diversity of interests.Our results show that the collab-orative similarity of both developers and projects negatively correlates with degree in watching layer,while a positive correlations is observed for developers in forking layer and no obvious correlation is observed for projects in forking layer.The reason behind this result is the different usage of these social functionalities provided by GitHub.Further we study developers' interests from the perspective of profession levels and propose a new index called dependence centrality similarity.Our results show that the nearest neighbors' centrality is negatively correlated with degree for developers.However,for projects,the distribution of nearest neighbors' centrality fluctuates around the average,indicating that as a developer keeps focusing more projects,he/she tends to be more pragmatic and pays more attentions to projects suitable for his own work but not those of high-profession and as a project attracts more attentions,it attracts both high-profession and ordinary developers with a balanced ratio.And obvious negative correlations between dependence centrality similarity and degree are observed for both developers and projects of both watching and forking,indicating that the diversity of profession levels increases when degree increases.