Advanced Techniques And Algorithms To Collect, Analyze And Visualize Spatiotemporal Data From Social Media Feeds

Abstract

During the past decade, the social media evolution has grown to become an essential part of our lives, producing an incredible amount of user-generated and voluntarily contributed content, about people's everyday activities, thoughts and interests.Besides the intent of the users and the platforms, we believe that the information residing within these streaming datasets can be useful in other contexts as well, such as enriching collections and gazetteers, providing real-time notifications and alerts, identifying trends and providing academia and industry with valuable on-the-ground knowledge.To this end, this thesis focuses on exploring the potentials of these datasets by designing and implementing advanced algorithms and data structures to extract, analyze and visualize data and information from the social media traffic. Throughout this dissertation we describe several cases of using social media data to solve different problems, or enhance existing solutions.The cases studied include: (i) locating and describing geospatial features, (ii) integrating geospatial content from several sources, (iii) providing multidimensional interactive and interconnected summaries of social media traffic, (iv) enhancing event detection by suggesting a methodology to rank events based on their spatial footprint, and (v) linking together places using people's activities.Our extensive work is accompanied by a series of experiments and examples. The lessons learnt allow us to generalize and modularize the workflow pattern discovered in all our previous efforts, which benefits any future work.