Dynamic volunteer assignment: Integrating skill diversity, task variability and volunteer preferences

Abstract

Non-profit organizations rely critically on volunteers for effective disaster response. Managing diverse skills and varying participation levels of volunteers poses significant challenges, especially under the fluctuating demands and the uncertainty of task completion typical of disaster scenarios. This study introduces a model that dynamically optimizes volunteer allocation, enhancing disaster response efficiency and volunteer engagement. Integrating a multi-task queuing model with a dynamic priority policy within a Markov Decision Process framework, the model aims to minimize costs associated with task backlogs and volunteer services. Utilizing deep neural networks and policy iteration, the model handles large-scale environments and reduces costs through volunteer allocation. This adaptive approach responds to changing task demands, focusing on minimizing the long-term operational costs of volunteer management. Experimental results demonstrate that this dynamic allocation significantly reduces disaster response costs and decreases volunteer participation expenses without requiring additional resources, underscoring the importance for non-profit organizations to strategically manage their volunteer labor, taking into account the attributes of volunteers.