Affiliations: [a] Department of Computer Science and Engineering, Kno.e.sis Research Center, Wright State University, Dayton, OH, USA. E-mail: derek.doran@wright.edu | [b] Department of Computer Science and Engineering, University of Connecticut, Storrs, CT, USA. E-mails: karl.severin@engr.uconn.edu, ssg@engr.uconn.edu | [c] Industrial Software Systems, Data Analytics Group, ABB Corporate Research, Raleigh, NC, USA. E-mail: aldo.dagnino@us.abb.com
Correspondence:
[*]
Corresponding author: Derek Doran, Department of Computer Science and Engineering, Kno.e.sis Research Center, Wright State University, Dayton, OH 45435, USA. E-mail: derek.doran@wright.edu.
Abstract: Smart city initiatives rely on real-time measurements and data collected by a large number of heterogenous physical sensors deployed throughout a city. Physical sensors, however, are fraught with interoperability, dependability, management and political challenges. Furthermore, these sensors are unable to sense the opinions and emotional reactions of citizens that invariably impact smart city initiatives. Yet everyday, millions of dwellers and visitors of a city share their observations, thoughts, feelings and experiences, or in other words, their perceptions about their city through social media updates. This paper reasons why “human sensors”, namely, citizens that share information about their surroundings via social media can supplement, complement, or even replace the information measured by physical sensors. We present a methodology based on probabilistic language modeling to extract and visualize such perceptions that may be relevant to smart cities from social media updates. Using more than six million geo-tagged tweets collected over regions that feature widely varying geographical, social, cultural and political characteristics and tweet densities, we illustrate the potential of social media enabled human sensing to address diverse smart city challenges.
Keywords: Social media, smart cities, language modeling, geo-locations
