Integrating uncertain user-generated demand data when locating facilities for disaster response commodity distribution

This paper presents a new facility location problem variant with application in disaster relief. The problem is unique in that both verified data and unverified user-generated data are available for consideration during decision making. The problem is motivated by the recent need of integrating unverified social data (e.g., Twitter posts) with data from more traditional sources, such as on-the-ground assessments and aerial flyovers, to make optimal decisions during disaster relief. Integrating social data can enable identifying larger numbers of needs in shorter amounts of time, but because the information is unverified, some of it may be inaccurate. This paper seeks to provide a “proof of concept” illustrating how the unverified social data may be exploited. To do so, a framework for incorporating uncertain user-generated data when locating Points of Distribution (PODs) for disaster relief is presented. Then, three decision strategies that differ in how the uncertain data is considered are defined. Finally, the framework and decision strategies are demonstrated via a small computational study to illustrate the benefits user-generated data may afford across a variety of disaster scenarios.