| dc.description.abstract | Technological advancement enables the development of information and communication technology solutions (ICT from English Information and Communication Technology) from a simple niche, applied to a specific context, such as the computerization of macro environments, a city for example. There are several objectives that can be achieved for the latter case. To name a few: assistance with security features (cameras and sensors) and qualifying security environments (heat maps identifying risk zones). The present work presents the UFollower model (Ubiquitous Follower) which meets this scenario. The UFollower model proposes the use of ubiquitous computing to support security. It includes an ontology for people, resources and environments. Along with the ontology, a multiagent system is proposed to make inferences and support security. Using prediction mechanisms and historical contexts, it was possible to reach the rate of 70.35% and correct the inference of the occurrence of a crime. A differential contribution of the model is ubiquitous monitoring of the citizen, watching him in an ubiquitous way through data from the context to which he is inserted, as well as using information that the model deems useful according to the registered profile. The comparison with related works indicates that UFollower is the only proposal that presents the issue of Public Security, optimizing the action of agents through inferences made through Bayesian Networks. The evaluation is presented through processes defined in the form of experiments described in the scenario format, which aim to evaluate the hypotheses related to each one of them. The starting point for formulating each hypothesis is the fact that you have an idea of a cause and the effect related to it. A prototype was implemented and applied in an evaluation organized in three scenarios, where it was possible to prove the relevance of creating the model. | en |
