AbstractsSociology

Cities are complex systems, a particular dimension largely contributing to the complexity of cities is detected in the multiplicity of the behaviour of people who inhabit. In the urban context, these behaviours can be inferred from human activities. Understanding the distinctions of these activities between cities is crucial. Previous works carry out such studies using surveys, censuses, interviews and onsite observations, which are rather costly, requiring extensive human or other resources. With the development of Internet and information technology, the geo-referenced social media networks which link human activities to specific places have been more and more popular. It, therefore, becomes a good candidate for studying individual human activities in the city. In this work, we explore to what extent can social media data be used for studying differences of individual human activities in an urban environment, considering impacts of three aspects: the choice of social media platforms, the user roles and demographics, and the various of Point of Interests (PoIs). To perform this study, we design an experiment to compare user activities on multiple social media platforms in Rotterdam and Shenzhen. A pipeline for crawling and enriching Weibo data is developed and encapsulated as an extension to the Social-Glass framework. A novel dataset containing data from four platforms (i.e. Twitter, Instagram, Foursquare and Weibo) in two cities for two weeks is created and analysed at a disaggregate level across several dimensions. Results indicate that, in the context of user activity study, Twitter and Weibo shows distinct differences for resident regarding local, general, and neutral topics particularly for young and mid-age users on Twitter, and for teen and young users on Weibo. While, Instagram shows evident differences for non-residents concerning international, specific, daily and emotional topics, especially for young and mid-age users. In the meantime, various PoIs provide significant differences in user activity. Advisors/Committee Members: Bozzon, A..