Design and implementation of a visualization module for agent-based social simulations applied to radicalism spread

Abstract

Agent-based Social Simulation tools have been developed as one of the applications of agent technology. It constitutes the intersection of three scienti_c _elds, namely, agentbased computing, the social sciences, and computer simulation. Due to its increasingly use in social context, in this thesis, a web interface for its visualization is presented. The aim of this project is to develop a web application which allows the user to visualize and analyze the results of a simulation. It allows the user to analyze data in real time and launch new simulations which can be con_gured from the web. For this purpose, a model which tries to simulate the growth of radicalism in a society is developed. The model aims at improving the understanding of the inuence of social links on radicalism spread. The model consists of two main entities, a spread model and a network model. The network model updates the agent relationships based on proximity and homophily, it simulates information di_usion and updates the agents' beliefs. The model has been implemented in Python with the agent-based social simulator Soil and it has been evaluated using a sensitivity analysis while the application uses D3.js, which is a powerful JavaScript library, for rendering the results of the simulation and analyze them. This thesis is divided in three phases: the design and implementation of the visualization module, the analysis and research necessary to acquire the knowledge for modeling radicalism di_usion and the implementation of the model as well as making the application auto-con_gurable to simulate the model from it.