Intersection of machine learning and mobile crowdsourcing: a systematic topic-driven review

• Published in: Personal and ubiquitous computing 2025-02, Vol.29 (1), p.77-101
• Main Authors: Chen, Weisi, Hussain, Walayat, Al-Qudah, Islam, Al-Naymat, Ghazi, Zhang, Xu
• Format: Article
• Language: English
• Subjects: Big Data; Computer Science; Crowdsourcing; Data collection; Federated learning; Large language models; Machine learning; Mobile Computing; Original Paper; Personal Computing; User Interfaces and Human Computer Interaction
• ISSN: 1617-4909; 1617-4917
• DOI: 10.1007/s00779-024-01820-w

During the past decade of the big data era, mobile crowdsourcing has emerged as a popular research area, leveraging the collective intelligence and engagement of a vast number of individuals using their mobile devices. Another actively evolving area is machine learning, which has recently been augmented by the mobile crowdsourcing approach, especially for data collection and labeling. However, what happens when these two prevailing concepts meet? What topics have been discussed in recent literature? This paper adopts a systematic methodology, leveraging Latent Dirichlet allocation topic modeling for topic discovery from recent publications, to provide a comprehensive and insightful review of the intersection of machine learning and mobile crowdsourcing. Moreover, the paper highlights the emerging federated learning technology that integrates elements from both concepts. Key research questions are answered by examining discovered topics. The paper thoroughly discusses state-of-the-art developments and trends in combining these two concepts and explains the role of one concept in the other. The paper also addresses remaining challenges and outlines a future research agenda, including the potential incorporation of large language models into mobile crowdsourcing systems.