HUMAN MOBILITY IN URBAN SPACE

Nowadays we witness a rapid increase of people mobility as the world population has become more interconnected and is relying on faster transportation methods, simplified connections and shorter commuting times. Unveiling and understanding human mobility patterns have become a crucial issue to support decisions and prediction activities when managing the complexity of the today's social organization. The strict connections between human mobility patterns, the planning, deployment and management of a variety of public and commercial services have fueled the rise of a vast research activity. Throughout this work, we are more interested and mainly focusing on urban mobility because here most of the human interactions take place and mobility has the greatest impact on management and optimization of public and commercial services. In this thesis, we provided a general framework for dealing with the modeling importance of locations from a per-user perspective and identified a few novel properties of human mobility. Also through characterizing the transition patterns driving user movement among visited places, we pave the way to propose a new mobility model in urban spaces. Meanwhile relying on the relevance of visited places, we propose a new algorithm for detecting and distinguishing Home and Workplaces. And finally, we suggest a framework for predicting the different aspects of Encounter/Colocation events. By exploiting the weighted Bayesian predictor we could enhance the accuracy of prediction w.r.t. the standard naive Bayesian and also to some other state-of-the-art predictors.