Abstract

SALGADO, Ariel et al. Cohesion of the resulting structures in social networks: case study on the disappearance of people in the province of Tucumán. Inter disciplina [online]. 2021, vol.9, n.23, pp.83-96.  Epub May 05, 2021. ISSN 2448-5705.  https://doi.org/10.22201/ceiich.24485705e.2021.23.77347.

In this work, we propose a quality characterization of the structures arising in social networks, in terms of the homogeneity among the individuals belonging to those structures or groups. The proposed techniques are intended to be used in sparse networks, in which isolated groups of agents (clusters) emerge, but can be directly implemented for more dense networks, with communities are detected by any method. The group quality is quantified in terms of their inner similarity or homogeneity, based on a small set of TRUE/FALSE (belongs or not belongs to the category) categorical variables, known for all elements in the network, which were not used previously in the network definition, and also they are not necessarily disjoint. The methodology is to compare each group with what could be obtained if the connections were randomly assigned, to quantify each category’s presence in that group, but also taking into account that categories should not mix too much if they do not mix in the whole population. The specific application of these methodologies aims to take advantage of the network characterization, helping to make priority rankings in the investigation of missing people, because it gives connections, bonds, paths and nodes that to help to understand the operational framework of the last Argentinian civic and military dictatorship, in the Tucumán province.

Keywords : networks; clusterization; inner similarity; external validation.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )