Speaker
Description
We study a social network where agents correspond to people, and links are relationships between agents. Each agent possesses a set of attributes. Distinguishing the signs of relationships between agents can be performed for each attribute separately or considering all attributes together. In the former case, we assume a simple edge is positive/negative when the two agents hold the same/different attribute. In the latter case, for a pair of agents $i$ and $j$, a normalized distance is calculated in the multidimensional space of attributes, $x_{ij}$, defining the multi-edge link as follows: a link is positive when $x_{ij}\leq\Theta$, and it is negative otherwise where $\Theta$ is a threshold parameter. We apply our signed network construction definition to study the NetSense dataset, which contains data about relationships between university students and their opinions on important social topics. We construct simple and multidimensional triads and test for which condition Structural Balance Theory (SBT) principles can be measured in the system, i.e. ”friend of my friend or enemy of my enemy is my friend” etc. Density of balanced triads and triad transition probabilities are considered . Measures obtained for the real network are compared with those for three different null models and two randomized processes. Our results show that SBT influence is not observed in the case of simple edges. Triad densities for real networks are not statistically different from densities in null models. However, in the case of multi-edges, for the range of tolerance values, multidimensional triads are significantly more balanced in the real network. This means that structural balance dynamics are measurable only when considering multidimensional attributes.
