Author(s): Siqi Wei; Jian Yan; Qingyuan Yang; Zhe Chen
Linked Author(s):
Keywords: Flash Flood Mitigation; Knowledge Graph Construction; Multi-Modal Data Integration; Spatiotemporal Dynamics; Knowledge Representation Enhancement
Abstract: Smart water resource management is crucial for achieving high-quality water resource development. In the realm of smart water resources, the establishment of a knowledge platform built upon a knowledge graph is imperative. Flood disaster mitigation stands as a critical component within the domain of smart water resource management. To address the limitations of existing methods for constructing flood disaster knowledge graphs, which primarily suffer from single data sources and insufficient representation of spatiotemporal processes, this paper introduces a novel approach. It is rooted in the spatiotemporal dynamics of flood disasters and involves the creation of a comprehensive flood disaster representation model. Utilizing a multimodal neural network, this approach harmonizes the expression of remote sensing imagery and textual data. By way of illustration, we apply this approach to the 2023 flood disaster in Hebei Province, effectively integrating data from diverse, heterogeneous sources. This process facilitates the transformation of multimodal data into actionable knowledge, ultimately resulting in the construction of a sophisticated flood disaster knowledge graph. The outcomes of this methodology reveal a significant improvement in the holistic representation of flood disaster events, offering valuable decision support for flood disaster emergency management.
Year: 2024