Exploring the spatiotemporal patterns of urban characteristics using social big data and GeoAI

Keywords: urban characteristic, spatial big data, geotagged social-sensing image, GeoAI
Session Type: Virtual Paper Abstract
Day: Saturday
Session Start / End Time: 2/26/2022 05:20 PM (Eastern Time (US & Canada)) - 2/26/2022 06:40 PM (Eastern Time (US & Canada))
Room: Virtual 1

Abstract

This study is to investigate the types of urban clusters that can be found within a city and, in particular, spatially identify cultural characteristics within the clustered area by the cutting-edge geospatial artificial intelligence (GeoAI) methodologies. Analyzing citizens’ interests helps predict the urban dynamics where people live. For this reason, we have been using social media data that is nowadays called big data. For example, geotagged social-sensing images acquired from an online platform contain a large amount of meaningful information such as people’s interests, their activities, and cultural or physical characteristics that occur in the city. It can eventually show how people are connected to the cultural and physical environments of the city. Hence, given that big data is used in real-world research projects, geographers can spatially analyze how the cultural and natural environments are connected with the people. This research used Flickr images collected from 2015 to 2021 in which it spatially characterizes cultural characteristics in the urban area. As for the methodology, a hierarchical density-based spatial clustering of application with noise (HDBSCAN) has been used to determine the types of spatial patterns of urban clusters. In addition, a convolutional neural network (CNN) based deep learning model was employed to classify the cultural characteristics in each of the urban clusters. Therefore, this study is expected to be helpful in understanding the heterogeneous patterns of the recent city environment.