Abstract:
Place is defined by physical, social, and economic activities and processes. Understand- ing the complexity of socially constructed places is a fundamental question in geography, sociology, and many other social sciences. Meanwhile, the growing amount of user volunteered geographic information (VGI) leads us to study place through a new perspective. For instance, Flickr users report local activities in various geographic locations that capture individualistic experiences and impressions of the locations. Many previous studies utilizing non-textual VGI have focused primarily on analyzing geographical footprints of places, which separated place from its meaning. This dissertation argues that the textual part of VGI provides us with unprecedented opportunities for deriving patterns of place meanings on an individual level. More specifically, three research questions are pursued in this dissertation. First, how to quantify placeness (i.e., place identities) that has been traditionally studied via theoretical and qualitative methods? Second, as place being innately interconnected, how can we assess connections between places in networks so that we can apply network science to analyze complex connections between places? Third, as geo-textual data can also reveal social events, how to trace critical events across places using geo-textual data? In order to answer these research questions, this dissertation leverages advances in machine learning, natural language processing and network analysis techniques on geo- textual data. By doing so this dissertation is able to build foundations for geo-textual data analytics and thus providing a new lens to study places and the connections between them from the bottom up. Overall, this dissertation showcases an interdisciplinary effort in computational social science research that combines computational textual data analytics and social scientific theories including human geography and sociology.
