inContexto: framework to obtain people context using wearable sensors and social network sites

• Autores: Gonzalo Blázquez Gil
• Directores de la Tesis: José Manuel Molina López (dir. tes.) , Antonio Berlanga de Jesús (dir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2015
• Idioma: inglés
• Tribunal Calificador de la Tesis: Juan Luis Pavón Mestras (presid.) , Miguel Ángel Patricio Guisado (secret.) , Nayat Sánchez-Pi (voc.)
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • Ambient Intelligent (AmI) technology is developing fast and will promote a new generation of applications with some characteristics in the area of context awareness, anticipatory behavior, home security, monitoring, Health Care and video surveillance. AmI Environments should be surrounded by multiples sensors in order to discover people needs. These kind of scenarios are characterized by intelligent environments, which are able to recognize inconspicuously the presence of individuals and react to their needs. In such systems, people are conceived as the main actor, always in control, playing multiple roles, and this is perhaps the new real facet of research related to AmI: it introduces a new dimension creating synergies between the user and the environment. The AmI paradigm sets the principles to design pervasive and transparent infrastructures being capable of observing people without prying into their lives, and also adapting to their needs. There are several basis concepts to consider for retrieving people context, however the most important for users is that sensors devices must be unobtrusive. Many technologies are conceived as hand-held or wearable, taking advantage of the intelligence embedded in the environment. Mobile technologies and Social Network Sites make it possible to collect people information anywhere at anytime, and provide users with up-to-date information ready for decision-making processes. Nevertheless, the management of these sensors for collecting user context poses several challenges. Besides the limited computational capabilities of mobile devices, mobile systems face specific problems that cannot be solved by traditional knowledge management methodologies and tools, and thus require creative new solutions. This dissertation proposes a set of techniques, interfaces and algorithms for the implementation of inferring context information from new kind of sensors (Smartphones and Social Networking). The huge potential of both new sensors have motivated us to design a framework that can intelligently capture different sensory data in real-time. Smartphones may obtain and process physical phenomena from embedded sensors (Accelerometer, gyroscope, compass, magnetometer, proximity sensor, light sensor, GPS, etc.) and SNS the affective ones. Subsequently this information could be transmitted to remote locations without any human intervention. The mechanisms proposed here are based on the implementation of a basic framework that modifies information from the raw data to the most descriptive action. To this end, the development of this thesis starts from a inContexto framework which exploits off-the-shelf sensor-enabled mobile phones and SNS people presence to automatically infer people’s context. The main goals of our architecture are: (i) Collection, storage, analyse, and sharing of the user context information, (ii) Plug-and-play support for a wide variety of sensing devices, (iii) Privacy preservation of individuals sharing their data, and (iv) Easy application development. Furthermore our inContexto has been implemented to allow third party application to participate and improve people context.