BDD: Human-Centered Situational Awareness Platform for Disaster Response and Recovery

有效和全面的数据收集及其及时分析对于任何灾害应对和恢复系统都至关重要，以便在灾害期间挽救人们的生命。然而，获取灾区的全面数据并对其进行快速分析以将数据转化为可操作的知识是数据科学的主要挑战。此外，收集到的知识的有效呈现给人类决策者是一个开放的问题。 在这个项目中，这些数据收集、分析和呈现的挑战被统称为“情境意识”，并由来自美国南加州大学（USC）和日本国立信息学研究所（NII）的数据科学家组成的合作团队进行研究。这个项目的第一个目标是设计有效的技术来进行全面的数据收集。挑战在于，在灾难期间，信息的可用性存在空间偏差，并且可用信息没有很好地覆盖某些地区。为此，南加州大学的空间众包平台，被称为MediaQ，被用来收集图片和视频点播从移动的设备的人在灾区附近，以促进有效的收集，编排和汇总信息。第二个目标是开发有效的方法来组织和分析人类决策者的传入数据流，考虑到其准确性未知的非结构化数据的数量和种类的挑战。美日合作团队利用他们在开发空间-时间-主题分析引擎和地理空间图像过滤工具方面的综合专业知识来实现这一目标。最后，为了向人类利益相关者有效地展示知识，NII的DiVE虚拟环境引擎用于促进现场（现场人员）和非现场（中心的灾难管理人员）利益相关者的无缝展示和沟通。

Efficient and thorough data collection and its timely analysis are critical to any disaster response and recovery system in order to save people's lives during disasters. However, access to comprehensive data in disaster areas and their quick analysis to transform the data to actionable knowledge are major data science challenges. Moreover, the effective presentation of the collected knowledge to human decision-makers is an open problem. In this project, these challenges of data collection, analysis and presentation are collectively referred to as 'Situational Awareness', and are studied by a collaborative team of data scientists from the University of Southern California (USC) in United States and the National Institute of Informatics (NII) in Japan.The first objective of this project is to devise effective techniques for comprehensive data collection. The challenge is that during disaster the availability of information is spatially biased and some areas are not well covered by available information. Towards this end, USC's spatial crowdsourcing platform, dubbed MediaQ, is utilized to collect pictures and videos on-demand from mobile devices of people in the vicinity of the disaster areas to facilitate effective collection, orchestration and aggregation of information. The second objective is to develop efficient methods for organizing and analyzing incoming data streams for human decision makers, given the challenges of volume and variety of unstructured data whose veracity is unknown. The collaborative US-Japan team takes advantage of their combined expertise in developing Spatial-Temporal-Thematic analytics engines and Geospatial Image Filtering Tools to meet this objective. Finally, for an effective presentation of the knowledge to human stakeholders, NII's DiVE virtual environment engine is used to facilitate seamless presentation and communication for both onsite (humans in the field) and offsite (disaster managers at the center) stakeholders.