Resumo:
This work addresses the challenge of volunteer allocation in crisis scenarios, such as
natural disasters, which demand agile, efficient, and adaptable management systems. To
tackle this problem, a task allocation engine was developed and integrated into Contask,
a prototype system designed for context-aware task distribution. The allocation engine
is based on Integer Linear Programming (ILP), combined with heuristics, with the goal
of optimizing the distribution of microtasks in an efficient and context-sensitive manner.
The proposal considers multiple criteria including availability, skills, location, and task
priority drawing inspiration from existing microtasking models. The choice of combining
ILP with heuristics was grounded in a systematic literature review, along with a comparative analysis and an evaluation of the model’s applicability to the proposed scenario.
This process revealed that alternative approaches present practical limitations related to
solution stability and computational cost, especially in emergency contexts where rapid
and robust responses are required. The requirements elicitation was conducted through
an exploratory study with 15 participants, including experienced volunteers and managers
of social initiatives, using questionnaires and thematic analysis to identify key challenges and expected functionalities. Based on these requirements, Contask was redesigned
with the implementation of the allocation engine, resulting in an updated prototype of
the solution. This prototype was initially evaluated through usability tests and a focus
group with six managers, including members of the Mãos que Ajudam project. Subsequently, controlled simulations representing critical situations were performed to further
assess system performance. The results indicate strong acceptance of the tool and a preference for automatic allocation, suggesting potential practical applicability and meaningful
contributions to intelligent volunteer management in emergency situations.
