CHS: Medium: Hyperlocal and Hypertemporal Information in Mass Emergencies Events: Next Generation Crisis Informatics Data Collection and Analytics

该项目将通过将为基于社交媒体的危机信息的后处理而开发的广泛分析转移到数据收集点来推进危机信息学的实证研究，以创建一套智能数据收集技术。 危机信息学是一个多学科领域结合人机交互和社会计算;其中心宗旨是人们使用个人信息和通信技术，以创造性的方式应对灾害，以科普不确定性。这项研究将揭示人们如何应对灾难，无论是集体还是个人，通过检查他们的数字痕迹。这将有助于制定更好的应急管理做法和政策。该研究还将通过利用灾害期间产生的大量社交媒体数据集，促进人们在灾害中所需信息的传递。过去的研究表明，当地的人在网上寻找和提供信息，以帮助自己和彼此，而且这些信息是高度本地化和时间性的，当与全球对话相比，否则会出现。目前从社交媒体流中获取有用信息的技术是有限的，因为它们通常无法适应社交媒体制作的巨大增长;它们往往仅限于检查社交媒体帖子的语言，而不是其元数据;社交媒体的收集方式往往禁止深入探索和重新使用数据所需的分析。这项研究依赖于调查人员对灾难中社交媒体行为的性质进行的实证研究，并在社交媒体流的数据收集点上采用了严格的演绎和归纳分析方法。这将实现智能数据收集，可以执行实时分析，对通过社交媒体传播的信息的超时空和超本地特征进行分析。凭借敏捷性和严谨性，本研究中开发的解决方案将能够追求多个紧急关注的线程，这些线程将产生多个同时的数据收集轨迹。这将更好地实现对灾害事件发生期间社交媒体记录中发生的事情的实时分析和可视化。

This project will advance the empirical study of crisis informatics by moving the extensive analytics developed for post processing of social media-based crisis information to the point of data collection to create a set of intelligent data collection techniques. Crisis informatics is a multidisciplinary field combining human-computer interaction and social computing; its central tenet is that people use personal information and communication technology to respond to disaster in creative ways to cope with uncertainty. This research will reveal how people respond in disasters, both collectively and individually, by examining their digital traces. This will help create better emergency management practice and policy. The research will also advance the delivery of the information people need in disasters by leveraging the vast social media data sets that are produced during them. The safety of people and provision of emergency care as a function of improved situational awareness are the direct benefits of conducting this research.Past research shows that people local to an event seek and provide information online to help themselves and each other, and that this information is highly localized and temporalized when compared to the global conversation that otherwise arises. Current techniques to derive useful information from the social media stream are limited because they usually cannot scale to the vast growth in social media production; they are often restricted to examining only the language of social media posts and not their metadata; and social media are often collected in ways that prohibit the kinds of analyses that are needed for deep exploration and re-use of the data. This research relies on the body of work that the investigators have empirically conducted on the nature of social media behavior in disasters, and incorporates those rigorous deductive and inductive analytical methods at the point of data collection of the social media streams. This will enable intelligent data collection that can perform real-time analysis that bore in on the hypertemporal and hyperlocal features of information communicated via social media. With agility and rigor, solutions developed in this research will be able to pursue multiple threads of emergent concern that will spawn multiple simultaneous data collection trajectories. This will better enable real-time analysis and visualization of what is happening in the social media record during disaster events as they unfold.