Gómez Carmona, OihaneLópez de Ipiña González de Artaza, DiegoPuerta Beldarrain, Maite2026-02-172026-02-172025-09-25https://hdl.handle.net/20.500.14454/5130Citizen Science (CS) has become an essential approach for addressing real-world challenges by actively involving non-professionals in scientific research. Many CS initiatives rely on voluntary contributions rather than financial incentives, making participant engagement and task allocation critical challenges. One of the most complex aspects within CS is the field of Spatial Crowdsourcing (SC), where tasks must be performed in specific geographic locations within a limited timeframe. Traditional SC solutions typically depend on monetary rewards to ensure participation, which is not generally suitable for CS, where motivation is often driven by personal interest, learning, or altruism. For this reason, this dissertation explores Human-AI Collaboration as a mechanism to enhance task allocation in CS, proposing that collaboration between humans and AI-based systems can balance efficient task completion with a positive user perception, fostering greater acceptance and adoption of Spatial Crowdsourcing solutions without relying on financial rewards. For validating this hypothesis, this research develops a collaborative task allocation strategy, where a spatial task allocation recommender dynamically suggests spatial execution of tasks in the most needed locations while taking into account user preferences, their behavioural patterns, and availability. For that this dissertation introduce three task allocation strategies - Volunteer Task Allocation Strategy (VTAS), Bio-inspired VTAS (Bio-VTAS), and Hybrid-VTAS - which integrate AI-driven recommendations with an active user involvement in the decision-making process. That is, rather than simply automating task distribution, these systems engage volunteers as collaborative agents, enabling them to influence and adapt AI-driven recommendations based on their own motivations and constraints. By shifting from a purely automated system to a cooperative framework, this approach envisions to increase engagement, optimize task execution, and maintain equitable workload distribution across all locations and participants. The core of this dissertation has been to design and evaluate three alternative spatial crowdsourcing strategies which incrementally satisfy three design requirements, thus leading towards an optimal approach for an effective and user-aware spatial task allocation recommender. To assess the effectiveness of these spatial crowdsourcing strategies, the research combines simulated experiments with real-world case studies conducted at the University of Deusto. The evaluation examines different conditions, including variations in crowd size, task availability probabilities, and task completion probability. Results, against baseline algorithms which suggest nearest task locations to users, demonstrate that the proposed AI-assisted spatial task allocation recommender not only improves task completion rates and optimizes spatial distribution but also enhances user trust, motivation, and long-term engagement. Feedback from participants was collected from the real-world case studies, both in the form of structured questionnaires and semi-structured interviews. The obtained insight highlights the importance of transparency and control in AI-driven recommendations, reinforcing the idea that collaboration-rather than mere automation-plays a key role in sustaining volunteer participation in CS. In summary, this research contributes to the advancement of AIdriven crowdsourcing, human-centred computing, and Citizen Science paradigm by offering a scalable and incentive-free approach to task allocation. The findings suggest that collaboration between humans and AI can create more sustainable, engaging, and effective Citizen Science initiatives, bridging the gap between efficiency and human motivation.engCiencias TecnológicasTecnología de los ordenadoresUnidades centrales de procesodoctoral thesis