Human-Centered Innovative Technology for Resilient Built Environment
M.E. Rinker, Sr. School of Construction Management University of Florida
Embedded Files
Crisis Response and Disaster Resilience
Crisis Response and Disaster Resilience
Project summary
Project summary
Utilizing big data extracted from social media and crowdsourcing platforms, this project aims to optimize solutions for crisis response and build long-term resilience against disaster. This research contributes to the body of knowledge of big data applications in disaster science and crisis response, and development of new engineering models to support decision-making in the disaster context.
Summary of project
1. Damage Assessment with DIRR & DARR
1. Damage Assessment with DIRR & DARR
Research goal
This research aims at exploring the feasibility of using crowdsourcing data to identify the critical affected areas during disasters.
Research method
Utilizing unsupervised machine learning method for identifying disaster- and damage-related social media data, this study calculates the social media indexes including disaster-/damage-related ratio (DIRR/DARR) and sentiment. By conducting the correlation analysis between social media indexes and hurricane damage data (i.e., damage rate), this research demonstrates the feasibility of using social media data to support rapid damage assessment (Yuan and Liu 2018a).
Research results
By the correlation analysis and comparative analysis of geographic distribution between Twitter data and claims data, this research demonstrate a close relationship between social media data and claims data. We conclude that using social media data to identify the critical affected areas during Hurricane Matthew to support rapid damage assessment is feasible. The results can benefit crisis response managers in situation awareness.
Damage rate: the ratio between the number of insurance claims and the population in a county. (Yuan and Liu 2018a)
DIRR: the ratio between the number of disaster-related tweets and the number of general tweets in a county. (Yuan and Liu 2018a)
2. Damage Assessment with DARR & Sentiment
2. Damage Assessment with DARR & Sentiment
Research goal
This research aims at exploring the feasibility of using social media data to estimate damages during disasters.
Research method
This research firstly proposes to use supervised machine learning approach to identify the damage-related social media data. Secondly, it analyzes users’ tweet frequencies and introduces the annual average sentiment as the baseline to calculate the normalized sentiment. Correlation analysis is conducted between social media index and the insurance claim data in Hurricane Matthew.
Research results
The results show a strong and positive correlation between the DARR and claim data. A strong and negative correlation is found between sentiment and claim data. The adjusted R-square of final regression model between damage and social media index demonstrates the feasibility of mining social media data for rapid damage assessment.
DARR: the ratio between the number of damage-related tweets and the number of general tweets in a county (Yuan and Liu 2019)
The normalized sentiment during Hurricane Matthew for counties of Florida (Yuan and Liu 2019)
3. Forensic Disaster Investigations
3. Forensic Disaster Investigations
Research goal
This research aims at investigating causes of damages in Hurricane Harvey through the analysis of hazard, exposure and vulnerability.
Research method
‘Build back better after disaster’ is proposed in UNDRR’s Sendai Framework for Disaster Risk Reduction (SFDRR). A critical prerequisite of risk prevention measures for natural hazards is from the results of forensic disaster investigations (FDIs). Using scenario analysis approach, this research investigates the causes of Hurricane Harvey impacts from the analysis of hazard, exposure and vulnerability.
Research result
This research summarizes the causes of damages to exposed evacuees by analysis of variables (i.e., hazard, exposure and vulnerability) from individual scale to aggregated scale. The results provides urban planners with a new data source and data analytic method to identify the critical factors contributing to disaster damages and build a more resilient future (Yuan and Liu 2018b).
Hazard map: Each triangular represents an evacuee's location in flood Harvey. Water depth is considered as hazard factor. (Yuan and Liu 2018b)
Evacuees' vulnerability map: This research mainly considers age and health conditions for evacuees' vulnerability analysis. (Yuan and Liu 2018b)
References:
References:
Yuan, F., Liu, R. (2020). Mining social media data for rapid damage assessment during Hurricane Matthew: a feasibility study. Journal of Computing in Civil Engineering, 34 (3), 05020001.
Yuan, F., Liu, R. (2019). Identifying Damage-Related Social Media Data during Hurricane Matthew: A Machine Learning Approach. Proc., Computing in Civil Engineering 2019, Atlanta, GA, US.
Yuan, F., Liu, R. (2018a). Feasibility study of using crowdsourcing to identify critical affected areas for rapid damage assessment: Hurricane Matthew case study. International journal of disaster risk reduction, 28, 758-767.
Yuan F., Liu, R. (2018b). Crowdsourcing for forensic disaster investigations: Hurricane Harvey case study. Natural Hazards, 93(3), 1529-1546.
Yuan, F., Liu, R. (2018c). Integration of social media and unmanned aerial vehicles (UAVs) for rapid damage assessment in Hurricane Matthew. Proc., 2018 Construction Research Congress. New Orleans, LA, US.
Yuan, F., Liu, R., Mejri, O. (2017). An information system for critical infrastructure damage assessment supporting forensic analysis: a case study in Fort McMurray. Proc., International Conference on Sustainable Infrastructure. New York, NY, US.
All articles and conference papers published at the Lab are also available through this website.
Visit the Rui Liu's Lab page in researchgate for more information.
Visit the Rui Liu's Lab page in researchgate for more information.
Google Sites
Report abuse