Online profiling of volunteers in public health emergencies: insights from COVID-19 in China

Abstract

Background: During public health emergencies, the diverse backgrounds of volunteers pose numerous management challenges. This study aims to develop an online profiling model of volunteers using social media data to achieve a more comprehensive and objective understanding of them.

Methods: In the proposed model, the study designed five profiling tags: basic information, sentiment, topic features, interest preferences, and online social engagement. K-Modes clustering was employed to implement the profiling. To validate the feasibility of the model, an empirical study was conducted using Weibo data from 1,070 volunteers during the COVID-19 pandemic in China, resulting in the online profiling of these volunteers.

Results: Four categories of volunteers could be identified: Public Affairs Pioneers (32.4%), Diary Record Lurkers (32.8%), Social Topic Sharers (20.9%), and Fashion and Entertainment Influencers (13.9%). Overall, volunteers were predominantly female, generally interested in entertainment, relatively satisfied with their volunteer work, and possessed a sense of social responsibility. The four categories of volunteers exhibited distinct characteristics in terms of interests, online social behavior, and influence.

Conclusions: The proposed online profiling model objectively captures the characteristics of volunteers during public health emergencies. The four volunteer categories identified through the empirical results provide a multidimensional and comprehensive understanding of volunteers. For different volunteer categories, official agencies can tailor their recruitment, management, and training strategies to better suit the specific needs and strengths of the volunteers, thereby enhancing the effectiveness and efficiency of volunteer engagement and ensuring volunteers are well-prepared and supported in their roles.

Keywords: COVID-19; Online profiling; Public health emergencies; Social media; Volunteers; Weibo.

Fig. 1

The construction process of the online profiling model. Notes: Data Collection and Processing corresponds to Step 1, while Online Profiling Model Construction includes Steps 2 and 3. Specifically, Tag Design corresponds to Step 2, and Profiling Implementation corresponds to Step 3. Finally, Result Analysis corresponds to Step 4

Fig. 2

Online social engagement indicators

Fig. 3

Distribution of sentiment tags

Fig. 4

LDA topic modeling bubble chart

Fig. 5

Distribution of interest preferences tags

Fig. 6

Cost function curve

Fig. 7

Tag word cloud

Fig. 8

Radar chart of categories of interest

Fig. 9

Topic heat map