Discovering A Collective Sense of Place Through Crowd-Generated Content




Jenkins, Andrew

Journal Title

Journal ISSN

Volume Title



Place is generally defined as location given meaning through human experience. The topic of place has been widely debated and studied throughout geography and the social sciences as a theoretical construct. However, the rise and availability of user-generated content now affords new opportunities to computationally analyze and quantify the social meaning of place, but the question still remains of how well such content can be mined to discover place or so-called platial knowledge. This research investigates the question by focusing on the shared meaning of place by generalizing people’s collective sense of place. It is argued that taking a crowd-centric approach of collective and implicit sense of place meanings will lead to the discovery of emerging platial themes. Moreover, given the semantic-spatial-temporal characteristics of human activities within urban spaces, one can observe the emergence of unique themes that characterize different locations. In this dissertation, a novel quantitative approach is presented with statistical validation for deriving such platial themes from crowd-contributed content. This approach leverages unsupervised probabilistic topical n-gram modelling for dimensionality reduction, knowledge base labelling using semantic association, and spatial clustering with iterative distance analysis. Experimental results are presented from four different study areas that depict the emergence of unique places, thematic alignment across different data sources, and co-occurrence trends. The discovery and identification of locations that convey a collective sense of place contributes to the goal of observing how people transform a location to a place and shape its characteristics.



Geographic information science and geodesy, Computer science, Collective consensus, Place, Semantic, Spatial statistics