Detecção de comunidades no sistema de correio electrónico universitário [pdf]. David Rodrigues. ISCTE-IUL Instituto Universitário de Lisboa, Março 2009 (BibTex, RIS, Citeulike, PowerPoint)
The study of structured systems in social networks has gone through several developments by the use of graph theory in social sciences. On aspect that has been given considerable attention in recent years is the module or community detection in social networks. Several algorithms and strategies have been developed to identify the structure behind social interaction. Through a case study we show the existence of communities based on informal commu- nication that use the email system at ISCTE. We applied a set of hierarchical algorithms to detect communities. Also, we analyzed the hierarchical structure through the k-cores method and verified the transitivity of the communities detected through clique per- colation to put in evidence that informal communities are transversal to the institution departments. We also used a information variation measure to compare distances between different clusterings. We built a multi-agent simulation to model the informal communication mechanism of the email system, CIUCEU. This is used to verify the dependence of the system on the notion of social neighborhood, in the teachers network of ISCTE. We also analyzed the usage of real data and concluded on its implications of the sampling and drawing os multi-agent simulations.