Licentiatseminarium Zheng Ren
Välkommen att delta vid Licentiatseminarium för Zheng Ren inom ämnet Geospatial informationsvetenskap
Välkommen att delta vid Licentiatseminarium för Zheng Ren inom ämnet Geospatial informationsvetenskap
Titel: Living structure for understanding human activity patterns using multi-source geospatial big data
Opponent: Professor Itzhak Benenson, Tel Aviv University
Ordförande och examinator: Lektor Eva Salin, Högskolan i Gävle
Handledare: Professor Bin Jiang, Högskolan i Gävle och Professor Stefan Seipel, Högskolan i Gävle
Datum: 9 juni 2023
Tid: 10:00
Plats: Lilla Jadwiga, 12:108 och Zoom
Sammanfattning: Geographic space is not lifeless or neutral, but an intricate structure comprising numerous small and few large features in a recursive manner. Accurately understanding this structure is crucial to comprehending how it shapes human activities. Thanks to geospatial big data, researchers have unprecedented opportunities to study geographic space and human behaviours. This thesis leverages multisource geospatial big data, including tweets, OpenStreetMap, building footprints, and night-time light images, to understand the fundamental mechanisms that underlie geographic space. To overcome the limitations of conventional analytics in the big data era, we propose a topological representation and living structure based on Christopher Alexander's conception of space to gain better insights into human activities in geographic space.
We adopt scaling and topological analyses to reveal the underlying living structure of geographic space, which is established at different levels of scale in a nested manner. Our results demonstrate that tweet locations at different scales can be accurately predicted by the underlying living structure. We also investigate the reliability of social media users for predicting human activity and find that building footprint data and Twitter data show similar scaling patterns and statistical distributions, with high correlation and similarity. We propose an improved method for spatial clustering that is better suited for big data analysis. Finally, we use a topological approach to identify urban centres by fusing multiple sources of geospatial big data. The topological representation is a truly multiscale representation, and the living structure is an efficient and effective instrument for structuring geospatial big data to understand human activities both collectively and individually.
Välkommen!