Investigation of social media representation bias in disasters: Towards a systematic framework

Due to its real-time and human-centered nature, social media posts have been widely applied to provide rapid situational awareness in disasters, particularly from a human-centered perspective. To generalize social media-derived insights on a population, a pre-requisite is that the employed social media posts are capable of revealing the information of disaster-affected population without bias. Such wide application and pre-requisite underscore the importance of investigating social media bias for deriving reliable decision support insights in disaster management. However, a systematic framework that streamlines the investigation of social media representation bias is still missing. To address the research gap, we propose a framework comprising (1) the setting of an appropriate representation bias benchmark;(2) the modeling of the sampling uncertainty of social media-derived insights;and (3) the derivation and quantification of representation bias distribution across races/ethnicities. Public transit amid COVID-19 in the United States is studied for illustration purposes. Nation-level results show that the White group is over-represented, the Asian group is slightly over-represented, and the Hispanic and Black groups are under-represented throughout the studied period. The level of social media representation bias varies across the states of California, New York, Texas, and Florida, and it is inversely correlated with population ratios. Such findings are beneficial for decision-makers to use social media to derive reliable insights into disaster-affected population, thereby making informed operational decisions accordingly.

Public Opinions on COVID-19 Vaccines-A Spatiotemporal Perspective on Races and Topics Using a Bayesian-Based Method.

The COVID-19 pandemic has been sweeping across the United States of America since early 2020. The whole world was waiting for vaccination to end this pandemic. Since the approval of the first vaccine by the U.S. CDC on 9 November 2020, nearly 67.5% of the US population have been fully vaccinated by 10 July 2022. While quite successful in controlling the spreading of COVID-19, there were voices against vaccines. Therefore, this research utilizes geo-tweets and Bayesian-based method to investigate public opinions towards vaccines based on (1) the spatiotemporal changes in public engagement and public sentiment; (2) how the public engagement and sentiment react to different vaccine-related topics; (3) how various races behave differently. We connected the phenomenon observed to real-time and historical events. We found that in general the public is positive towards COVID-19 vaccines. Public sentiment positivity went up as more people were vaccinated. Public sentiment on specific topics varied in different periods. African Americans' sentiment toward vaccines was relatively lower than other races.

Public demand estimation following disasters through Bayesian-based information integration

Public demand estimation is essential to effective relief resource distribution following disasters. However, previous studies are limited mainly due to the strong complexity, dynamicity, and nonlinearity of public demand. This research proposes an innovative data-driven approach to use the sample information (i.e., social media and surveys) to estimate public demand for the entire population. Twitter-based demand percentage (TDP) is applied as the predictor of actual demand percentage, while survey-based demand percentage (SDP) is taken as the ground truth of the actual demand percentage. The sampling bias of social media users is removed through a systematic process that contains the prediction of social media users’ races/ethnicities and the weighted aggregation of demand percentages. The sampling uncertainty of TDP and SDP is modeled with a Bayesian-based approach that integrates prior knowledge and observations from social media and surveys. The relationship between TDP and SDP is learned through a polynomial model, which is used to estimate future actual demand percentage solely using TDP. To illustrate the feasibility and applicability of the research, we studied public demand for COVID-19 vaccines in the US. Results demonstrate that TDP is a strong predictor of actual demand percentage. This research fully makes use of the advantages of sample information, i.e., the near-real-time nature of social media and the high reliability of surveys, to achieve a reliable and rapid estimation of public demand of the entire population.