On the social and spatiotemporal aspects for urban computing

Abstract

In this thesis, we address urban sensing approaches, especially the challenge of providing scalable sensing systems of social, spatial and temporal variables for urban scenarios. We deal with the problem of obtaining adequate information to be used in the design of urban solutions and improve services sensitive to the social and spatiotemporal context. We investigate collective behavior in the spatiotemporal context through Online Social Networks (OSNs) due to their enormous popularity and acceptance among users of mobile personal devices. This acceptance provides significant user's engagement, where substantial amounts of data are shared online daily, resulting in massive repositories of contextual information. Initially, we investigate the spatiotemporal preferences of users in several cities around the world in a joint analysis with climatic data. The results showed that a subset of cities exhibits a dynamic behavior that concerns the users' spatial preferences, where temperature thresholds can characterize the shift. The preference shift includes places visited by users in the studied cities as well as the transition between regions of the city, such that this phenomenon may be observed in significant portions of the population and places. In this work, we also investigate the effects of spatiotemporal dynamics on the co-location of users. We propose a network model based on timely meetings to estimate the social graph based on the geographic proximity. The results show the changes in the structural characteristics of the graph over time. Based on these insights, we propose a message forwarding protocol for delay-tolerant applications capable of increasing delivery rate while optimizing message replication and delay. In these investigations, we include the spatiotemporal dynamics of urban areas regarding content consumption. We jointly investigate the location of users and their content of interest according to the metadata used in this work provided by the OSN. The results show that the studied areas can provide substantial demands for redundant content in small spatiotemporal windows, such that users could cooperate to provide content locally and offload the demand. From these observations, we propose a distributed cache management mechanism able to take advantage of users' social and spatial persistence. Finally, we combine OSN data and official local government data to formulate an urban sensing framework capable of assessing characteristics of urbanism, sociability, and mobility of citizens of a city. In this work, we propose methodologies of sensing and analysis of social and spatiotemporal characteristics in urban settings. The proposed methods and applications favor the use of publicly available data to provide effective generalization capability. Thus, the applications and contributions of this work can be reproduced in cities not contemplated in this study, and other cities can take advantage of urban sensing.