KANDOR — Knowledge Analysis of Neighborhood Dynamics and Online Relationships

Abstract

With the emergence of smartphones and location-based social networks, a large amount of user-generated data has become available to better understand city dynamics and help urban planning. While most of the related works choose to focus on specific dimensions of the data, the proposed models aim to benefit from the extent of the information existing in social platforms. Thus, this paper explores the full potential of social media data and proposes novel clustering models for retrieving information about city dynamics and urban characterization by incorporating new dimensions to state of the art algorithms. Aspects such as venue rating, entropy and popularity may lead to new and more complete understanding of activities and trends in a city. Preliminary experiments show that it is possible to aggregate a large diversity of information from different social networks, and generate different and complementary visualizations of the city. Moreover, by applying these methods to urban environments, governments and citizens can better understand and build better sustainable cities together.