KE-CNN: A new social sensing method for extracting geographical attributes from text semantic features and its application in Wuhan, China

摘要: Abstract Social sensing is an analytical method to study the interaction between human and space through extracting reliable information from massive volunteered data. During ongoing COVID-19 pandemic, there are a large number of Internet social However, most them lack geographic attribute. In order resolve this problem, paper proposes convolutional neural network classification model based on keyword extraction synonym substitution (KE-CNN) which could determine attribute by semantic features text Besides, we realizes non-contact pandemic construct co-word complex capturing spatiotemporal behaviour people. Our research found that (1) mining can obtain public opinion events, (2) KE-CNN improves accuracy 5%–15% compared with traditional machine learning method. Through method, effectively establish medical, catering, railway station, education other types feature set, supplement missing spatial data tags, achieve good geographical seamless sensing.