Award Abstract # 1554412
CAREER: Information Accuracy and the Use of Social Data in Planning for Disaster Response

NSF Org: CMMI
Division of Civil, Mechanical, and Manufacturing Innovation
Recipient: UNIVERSITY OF ARKANSAS
Initial Amendment Date: January 29, 2016
Latest Amendment Date: September 23, 2022
Award Number: 1554412
Award Instrument: Standard Grant
Program Manager: Daan Liang
dliang@nsf.gov
 (703)292-2441
CMMI
 Division of Civil, Mechanical, and Manufacturing Innovation
ENG
 Directorate for Engineering
Start Date: February 1, 2016
End Date: January 31, 2023 (Estimated)
Total Intended Award Amount: $500,000.00
Total Awarded Amount to Date: $507,815.00
Funds Obligated to Date: FY 2016 = $467,758.00
FY 2020 = $40,056.00
History of Investigator:
  • Ashlea Milburn (Principal Investigator)
    ashlea@uark.edu
Recipient Sponsored Research Office: University of Arkansas
1125 W MAPLE ST STE 316
FAYETTEVILLE
AR  US  72701-3124
(479)575-3845
Sponsor Congressional District: 03
Primary Place of Performance: University of Arkansas
Fayetteville
AR  US  72701-1201
Primary Place of Performance
Congressional District:
03
Unique Entity Identifier (UEI): MECEHTM8DB17
Parent UEI:
NSF Program(s): CAREER: FACULTY EARLY CAR DEV,
GOALI-Grnt Opp Acad Lia wIndus,
HDBE-Humans, Disasters, and th
Primary Program Source: 01001617DB NSF RESEARCH & RELATED ACTIVIT
01002021DB NSF RESEARCH & RELATED ACTIVIT
Program Reference Code(s): 019Z, 041E, 042E, 043E, 1045, 1638, 9102, 9150
Program Element Code(s): 104500, 150400, 163800
Award Agency Code: 4900
Fund Agency Code: 4900
Assistance Listing Number(s): 47.041

ABSTRACT

The focus of this Faculty Early Career Development (CAREER) Program award is a new class of decision models capable of harnessing the power of uncertain social data for disaster response logistics planning. Information critical in planning logistics activities to support disaster response has traditionally been gathered via time-consuming efforts such as on the ground assessments. The use of social media during emergencies enables collecting a larger amount of potentially life saving information in a shorter amount of time. Many emergency managers have indicated their agency would take action on social data only after verifying it. This strategy contradicts the timeliness of social data; one of its primary advantages. The products of this research will directly address concerns over the usefulness of social data in decision making by quantifying the value of considering the information at various stages of verification. Results will be translated to the first responder community via a simulated game to provide a comparative demonstration of response planning with and without social data. New generations of engineers will be inspired to pursue careers in humanitarian logistics and infiltrate the field with social data concepts by integrating games and case studies into courses and summer programs and involving students in the research.

Novel models for uncertainty in real-time logistics planning will be developed that contribute to dynamic and stochastic routing in a number of ways. First, traditional assumptions of homeostatic probability distributions for modeled random variables do not hold, as crowdsourcing efforts constantly provide new information regarding the relative degree of belief in the accuracy of uncertain social data. Second, the models will allow timely action on uncertain requests instead of delaying resource allocation until the complete demand scenario is known. Sampling methods to account for these differences will be developed. Seminars and expert interviews with the first responder community will be conducted to select relevant routing problem variants and information formats. These activities will also determine a set of social data logistics strategies of practical interest to emergency managers, defined as policies that specify to what extent social data should be incorporated in a response plan. Developed models will be used to assess strategy performance across a diverse set of test instances based on real disasters. This will result in the identification of scenarios where social data integration can improve response efficacy, potentially transforming disaster response with methods that enable serving (saving) a larger number of needs (lives) in a shorter amount of time.

PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH

Note:  When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

(Showing: 1 - 10 of 13)
Emre Kirac and Ashlea Bennett Milburn "A general framework for assessing the value of social data for disaster response logistics planning" European Journal of Operational Research , v.269 , 2018 , p.486
Emre Kirac and Ashlea Bennett Milburn "A General Framework for Assessing the Value of Social Data for Disaster Response Logistics Planning" European Journal of Operational Research , v.269 , 2018 , p.486 10.1016/j.ejor.2018.02.011
Emre Kirac and Ashlea Bennett Milburn "A General Framework for Assessing the Value of Social Data for Disaster Response Logistics Planning" European Journal of Operational Research , v.269 , 2018 , p.486
Emre Kirac and Ashlea Bennett Milburn "A General Framework for Assessing the Value of Social Data for Disaster Response Logistics Planning" European Journal of Operations Research , v.269 , 2018 , p.486 https://doi.org/10.1016/j.ejor.2018.02.011
Erin Mullin and Ashlea Bennett Milburn "Disaster Response Routing with Social Data in Stylized Demand Scenarios" Proceedings of the 2018 IISE Annual Conference , 2018
Erin Mullin and Ashlea Bennett Milburn "Disaster Response Routing with Social Data in Stylized Demand Scenarios" Proceedings of the Industrial and Systems Engineering Annual Meeting , 2018
Erin Mullin and Ashlea Bennett Milburn "Disaster Response Routing with Social Data in Stylized Demand Scenarios" Proceedings of the Industrial and Systems Engineering Annual Meeting , 2018
Erin Mullin and Ashlea Bennett Milburn "Logistics to the rescue: an elementary introduction to planning in post-disaster decision environments" INFORMS Transactions on Education , v.21 , 2021 , p.152 10.1287/ited.2019.0234
Erin Mullin and Ashlea Bennett Milburn "Logistics to the Rescue: An Elementary Introduction to Planning in Post-Disaster Decision Environments" INFORMS Transactions on Education , v.21 , 2021 , p.152 10.1287/ited.2019.0234
Jannatul Shefa, Ashlea Bennett Milburn "Qualitative Analysis of Stakeholder Interviews to Explain Operational Characteristics of Staging Areas for Hurricane Disaster Response" Proceedings of the Industrial and Systems Engineering Annual Conference and Expo , 2023
Neel Chanchad and Ashlea Bennett Milburn "Impact of Replanning on A*-HOP Performance to Solve Stochastic CTP" Proceedings of the Institute of Industrial and Systems Engineers Annual Conference and Expo , 2023
(Showing: 1 - 10 of 13)

PROJECT OUTCOMES REPORT

Disclaimer

This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.

Logistics is key to minimizing suffering and loss of life after a disaster. The logistics management of response efforts is complicated by highly dynamic and uncertain decision environments. This CAREER Award focused on planning for disaster response in the face of uncertainty. Specifically, Dr. Milburn worked with stakeholders to identify the need for and develop decision-support models capable of considering uncertain user-generated information streams when allocating resources for disaster response.

Social platforms can play a critical role during disaster response, providing fast access to citizen-generated situational awareness (e.g., what roads are flooded). However, this information is initially not verified, and some of it may be inaccurate. This project supported the development of a general framework for investigating the value of acting on uncertain user-generated data prior to its verification in the context of disaster response. It has been demonstrated for facility location and routing problems arising in disaster response.

Expert interviews with disaster response logistics managers were conducted in order to gain insights into the practical considerations surrounding logistics operations conducted on networks disrupted by disasters. Interviews were transcribed and content analysis was used to identify themes. A majority of participants reported unexpectedly encountering disrupted roads during response operations, due for example to flooding or debris. When new disruptions are discovered, the information was inconsistently made available to other responders through official channels. Thus, a discrepancy can exist between the actual status of the road network and its official designation. These findings support the need for routing algorithms that are robust to undiscovered disruptions.

A model that addresses scenarios where resources must be routed through a disrupted network is the Canadian Traveler Problem (CTP). In it, an agent must travel through a disrupted graph from start to finish as quickly as possible, while only learning the status of individual edges upon arrival to nodes incident to the edge. This research develops and tests solution algorithms for three stochastic CTP variants. Each model has a single starting point, resembling the real-world practice of dispatching relief supplies from a Federal Staging Area (FSA).  The models differ according to the numbers of agents traveling the network and destinations.

(1)   Single-agent single-destination: A new solution approach is developed and tested. It uses a consensus function to select nodes to visit based on how frequently they appear in optimal paths across disruption scenarios. It finds new best-known solutions for approximately 1 in 3 test instances and provides better average case performance for 2 of 30 graphs tested.

(2)   Multiple-agent single-destination: In this problem, multiple agents travel towards a single destination and share road network information with each other. A new solution approach that combines path cost estimation and diversification techniques into an A*-based framework is developed and tested. The new planning policy for two vehicles often enables the discovery of shorter paths through disrupted networks than more sophisticated planning policies for one vehicle.  The approach improves upon the only existing multi-agent CTP computational study.

(3)   Multiple-agent multiple-destination: In this new problem, multiple agents at a single starting point travel to multiple distinct destinations and share information learned about the road network. Three solution approaches are developed and tested. Results indicate that when agents collaborate, path costs decrease by up to 7.5%. As the number of agents and destinations increase, the ability of communication to improve the quality of solutions increases.

The broader impacts of this work are multi-faceted.

1)    K-12 education: A game is developed for K-12 students that gives players the opportunity to develop routing plans for a disaster response scenario with uncertain social data inputs. The game was administered to participants in a total of 8 summer camps that enroll rising 5th-8th grade students, many from underrepresented groups in STEM.

2)    Higher education: Disaster response logistics materials are incorporated into courses in quantitative transportation logistics modeling.

3)    First responders: The first responder community participated in expert interviews that shaped this research. They benefit from the sharing out of information learned through the interviews. Further, a game that is similar to the K-12 game, but tailored to an adult first responder audience, is developed. Results suggest game participants' views of using social media data to support disaster response increases after playing the game.

4)    General public:  Project results suggest the advantages of acting on social data during disaster response efforts, even when there is the possibility of inaccuracy, outweigh the disadvantages of failing to include it in response plans. Results also suggest that decision-support models capable of considering probabilistic information about road network disruptions produce better relief supply routes than those that do not. While these insights stem from models that simplify reality in a number of ways, significant impacts to how disaster response is currently conducted in this country are anticipated as those assumptions are relaxed and the models approach real-world practice.

 


Last Modified: 05/31/2023
Modified by: Ashlea B Milburn

Please report errors in award information by writing to: awardsearch@nsf.gov.

Print this page

Back to Top of page