Clusters in networks with incomplete information: the disappeared in Argentina

CARIDI INÉS; DORSO CLAUDIO; GALLO PABLO; SOMIGLIANA CARLOS

Puerto Alegre, Brasil

Workshop; Physics Applied to Economics and social Sciences PAESS'08; 2008

Instituto de Física, UFRGS

Abstract de la presentación oral  (13 minutos): We describe a method developed in order to recognize communities in networks with lacking information. The method was conceived to be applied to the problem of people disappeared in the argentine province of Tucumán during the period 1975-1984. The information collected by the Argentine Forensic Anthropology Team (EAAF) is used to build a social network in which the missing are replaced by nodes and links stand for relations established according to different criteria reflecting characteristics of the nodes, as for example: political affiliation, date of disappearance, etc. The main hypothesis underlying this calculation is that strongly correlated subsets of nodes have ended up in the same CDC (Clandestine Detention Center) and afterwards they probably were disappeared (buried, cremated, etc.) in the same place. Through this calculations we aim to provide new, undetected, information to the EAAF in order to find the remains of the missing. We discuss different methods to validate our findings and to select the best set of parameters of the model. Encouraging results have already been obtained and will be shown in this presentation.