Using Natural Language Processing and Computer Vision to Measure Global Impacts of Extreme Weather Hazards

极端天气在全球范围内造成重大破坏和生命风险。该博士将开发强大的计算工具，以系统地观察基于社交媒体和在线新闻的极端天气的社会影响。这解决了缺乏社会影响观测来验证预测和改进基于影响的预警系统的突出问题。该研究分为三个实验问题，其总体目标是衡量全球范围内极端天气灾害的社会影响。1）自然语言处理（NLP）如何使用简短的社交媒体文本帮助识别极端天气事件的影响？2)NLP如何使用新闻文章中的长文本来监测全球极端天气事件？3)计算机视觉可以从图像和视频中确定危险影响吗？

Extreme weather causes major disruption and risk to life worldwide. This PhD will develop robust computational tools to systematically observe the social impacts of extreme weather based on social media and online news. This addresses an outstanding lack of social impact observations to validate forecasts and improve impact-based warning systems.The research is split into three experimental questions, with the overarching objective to measure the social impacts of extreme weather hazards at global scale.1) How can natural language processing (NLP) help characterise extreme weather event impacts using short-form social media text? 2) How can NLP monitor extreme weather events globally using long-form text from news articles? 3) Can computer vision determine hazard impacts from images and video?