Using social media data to construct and analyze knowledge graph for "7.20" Henan rainstorm flood event
Haipeng Lu, Shuliang Zhang, Yu Gao, Hengxu Jin, Pengcheng Zhao, Yixuan Gao, Yating Li, Wenxuan Wang, Yixuan Zhang
Abstract
Understanding the spatio-temporal evolution of rainstorm flood events is crucial for disaster prevention and mitigation. Social media data are highly beneficial for spatio-temporal analysis and timely perception of rainstorm flood events. However, effectively organizing and managing social media data remains an urgent problem, for which a knowledge graph is considered a feasible solution. This study proposes a framework for constructing a knowledge graph for rainstorm flood events using social media data, and analyzes the spatio-temporal evolution of these events based on it. Using the "7.20″ Henan rainstorm flood event as an example, we first conduct the cognition and representation of the rainstorm flood event, constructing an event knowledge representation model. The ERNIE-SVM model is then used for event classification, and knowledge is extracted using the BERT-BiLSTM-CRF hybrid model and the DRCC-DPA method. Finally, we integrate event knowledge and complete the construction of the knowledge graph, which is then used to analyze the spatio-temporal evolution of the events. The constructed knowledge graph achieves structure and content completeness scores of 0.79 and 0.81, respectively. The main purpose of this study is to combine the timeliness of social media data with the organizational benefits of knowledge graphs, enabling rapid spatio-temporal analysis of rainstorm flood events.