Prediction of Perception of Security Using Social Media Content

Abstract

The Perception of Security (PoS) refers to the opinion that persons have about security or insecurity in a place or situation. Real-time monitoring and the capacity of anticipation of citizen’s PoS are highly relevant for citizen’s security planning. Surveys represent the most widely used strategy to quantify PoS. Nevertheless, this approach cannot be applied continuously to obtain real-time monitoring or to predict future PoS. Recent evidence suggests that social network content may provide valuable information to quantify PoS. However, the prediction of these PoS quantifications remains poorly studied. We propose a novel strategy to quantify and anticipate PoS in short time windows using social network data. The model considers the external factors that may contribute to the publication of posts related to PoS and the retweeting phenomena. Results show that the proposed model may provide competitive predictive performances while keeps high levels of interpretability about the factors influencing PoS.

This work was funded by the project “Diseño y validación de modelos de analítica predictiva de fenómenos de seguridad y convivencia para la toma de decisiones en Bogotá”, at Bank of National Investment Programs and Projects, National Planning Department, Government of Colombia (BPIN: 2016000100036).