Two Informational Complexity Measures in Social Networks and Agent Communities

Most of the statistical social network analysis methods rely on a fixed network structure. Earlier methods of dyadic statistical analysis manage to provide quantification on degrees of mutuality between actors and triadic analysis does a step forward allowing validation of theories of balance and transitivity about specific components of a network. Each of these constitutes a subset of a more general k-sub graph analysis based on k-sub graph census extracted from the network architecture. Some sort of frequency analysis is done over these censuses from which probabilistic distributions can be evaluated. More recent single relational statistical analysis additionally allows the validation of statistic models through parametric estimation. Looking further for positional assumptions of groups of actors, stochastic block model analysis measure the statistical fitness of defined equivalent classes on the social network (Carrington, 2005) (Wasserman, 1994). All of these tools assume some sort a static network structure, a kind of snapshot of reality, over which statistical measuring is done and some degree of confidence is evaluated against real data. It is not of our knowledge any method of providing some sort of analysis on the dynamic structure of social networks in which the set of relations evolves over a certain amount of time. The method we propose has the purpose of allowing a probabilistic informational based evaluation of each actor’s inner ABSTRACT