The impact of a PEER funded research project "Text Analytics on Social Media for Resilience-Enabled Extreme Events Reconnaissance (TAR)" is highlighted below. The project Principal Investigator (PI) is Laurent El Ghaoui, Professor of Electrical Engineering and Computer Science, UC Berkeley. The research team includes Selim Günay, Project Scientist, UC Berkeley, Alicia Yi-Ting Tsai, Graduate Student Researcher, UC Berkeley, Chenglong Li, Graduate Student Researcher, UC Berkeley, and Minjune Hwang, Undergraduate, UC Berkeley.
The significant worldwide population growth and urbanization of the past century resulted in an era of global development and infrastructure construction on a massive scale, including buildings and other critical infrastructure systems. Recognizing the associated growth, the National Academy of Engineering has identified “restore and improve urban infrastructure” as one of the Engineering Grand Challenges of the 21st century. On a positive note, the major advancements made in sensor and communication technologies, artificial intelligence algorithms, and science-based understanding of natural hazards, taken in combination, provide a foundation for developing methods of advanced monitoring, maintenance, and reconnaissance of infrastructure. Moreover, the mass adoption of mobile internet-enabled devices, paired with wide-spread use of social media platforms for communication and coordination, has created new opportunities to better understand human responses to extreme events. These methods have the potential to tackle the above-mentioned grand challenge and achieve resilient communities following natural hazards. Being aware of the existing challenges and opportunities, this project presents the tools and methods aiming to achieve resilient communities through reconnaissance efforts. The project develops methods and software to collect news and social media posts after an extreme event to: a) create automatically generated new summaries for immediate report writing after an event, b) to extract key information, such as the recovery time, the most affected regions and infrastructure, and to relate these to the magnitude of the event, socio-economic consequences facing the community, etc. Application of this tool to several recent earthquakes are demonstrated and potential use of the tool along with extreme event reconnaissance networks can be further established.