灾害学2025，Vol.40Issue(4)：114-119,179,7.DOI:10.3969/j.issn.1000-811X.2025.04.017
DeepSeek应用于自然灾害风险机器学习预测研究的探索
Exploration of DeepSeek's Application in Machine Learning-Based Prediction Research for Natural Disaster Risks
摘要
Abstract
Against the backdrop of global climate warming,natural disasters are exhibiting new patterns of high frequency,sudden onset,and concurrent occurrence,making traditional prediction paradigms increasingly inadequate for urgent risk governance needs.Based on DeepSeek's intelligent interactive research design,se-mantic analysis with feature extraction,and AI-driven data governance coupled with knowledge discovery tech-nologies,a multidimensional analysis is conducted on 1,565 machine learning-based natural disaster risk predic-tion papers.The findings are as follows:①Technological dimension:China has maintained an absolute global lead over the past five years,with 2024 witnessing the fastest growth in machine learning applications;②Sce-nario dimension:urban disasters remain the core research focus,while emerging risks such as compound ex-treme events and cascading disasters are gaining prominence.Among leading journals,the International Journal of Disaster Risk Reduction dominates English publications,while Journal of Catastrophology leads Chinese pub-lications;③Methodological dimension:classification and regression models prevail,ensemble learning remains the dominant algorithm choice,deep learning demonstrates outstanding performance in spatiotemporal prediction and image processing,and multi-algorithm fusion has emerged as a cutting-edge trend.关键词
DeepSeek/人工智能/自然灾害风险预测/机器学习Key words
DeepSeek/artificial intelligence/natural disaster risk prediction/machine learning分类
资源环境引用本文复制引用
杨月巧,宁占金,李明媛,袁志祥,宋泽文..DeepSeek应用于自然灾害风险机器学习预测研究的探索[J].灾害学,2025,40(4):114-119,179,7.基金项目
大学生创新创业训练计划"人员密集场所的拥挤态势感知与辅助路径决策技术"(S202411775125) （S202411775125）
