In this chapter, a study on informal communication network formation in a university environment is presented. The teacher communication network is analyzed through community detection techniques. It is evident that informal communication is an important process that traverses the vertical hierarchical structure of departments and courses in a university environment. A multi-agent model of the case study is presented here, showing the implications of using real data as training sets for multi-agent-based simulations. The influence of the “social neighborhood,” as a mechanism to create assortative networks of contacts without full knowledge of the network, is discussed. It is shown that the radius of this social neighborhood has an effect on the outcome of the network structure and that in a university’s case this distance is relatively small.
Community detection can be very useful for performing an exploratory analysis of data, and its usage transverses several domains, from statistics to computer science, biology or psychology. In every science, it is necessary to deal with empirical data, and one of the first classifications that one tries is to group the data according to some property that might manifest itself similarly inside the groups. Several algorithms and techniques have been devised to accomplish this partitioning (Fortunato, 2010), but in practice all are faced with situations where a good partitioning isn’t accomplished, and new methods have to be devised. Some methods are robust, and can be used effectively to classify groups with sets of data that are very heterogeneous. On the other hand, some are very specific to certain problems and need initial conditions that are particular to make its results appropriate (Shortreed, 2006).
The span of techniques and algorithms that tackle the problem of classification and identification of communities in graph representations of data has seen a great amount of interest and developments in the past few years. The field isn’t confined only to traditional methods like graph partitioning, hierarchical clustering, partitional clustering and spectral clustering. There is a new set of divisive methods based on modularity, dynamical algorithms, spectral algorithms, and based on statistical inference that populated this field with several possible approaches on how to obtain information about structure (mainly social, but not limited to) that can be organized in the form of a graph (Fortunato, 2010).