Development of an agent-based model for simulating the influence of jihadist terrorist attacks on the 2004 Spanish general election

Gabriel Funes Gabaldón. (2022). Development of an agent-based model for simulating the influence of jihadist terrorist attacks on the 2004 Spanish general election. Final Career Project (TFG). Universidad Politécnica de Madrid, ETSI Telecomunicación.

Abstract:
18 years ago, one of the worst terrorist attacks that Spain has suffered overloaded with information the most relevant mass media channels. Most of this transmitted information was influenced by different political figures, including the government; and, in general, the messages received by the Spanish population discussed the responsibility of the attacks, pointing at two different perpetrators: ETA and Al-Qaeda. This controversy caused an unexpected turn on the general elections that took place three days after the explosions were produced, which resulted in a victory for PSOE even when a few days earlier, the most reliable surveys showed PP as the favorite of the candidacy. This project focuses on the design and implementation of an Agent-Based Social Simulation (ABSS) based on the development of an artificial model to simulate all Spanish voters who participated in the Spanish general elections of 2004. This model has been developed in the Soil and integrates two different entities: an external diffusion environment and an internal interaction network. The environment acts like the emitter of all messages to the population, generating an influence effect on the agents, as the media would do in voters. The internal environment is composed of the generation of an artificial network where neighbor agents interact between them, interchanging variables and parameters. The main goal of this project is to analyze and extract information related to the influence that mass media exercised in the population on the results of the general elections by talking about the terrorist attacks, all of which is extracted from the output files obtained after running the simulation several times. We achieve this by filtering the data and performing an analysis of sensibility, a comparison between the generated models, and visualization of the most important variables and their evolution through simulation.