Federated Dynamic Match-Making for Co-opetition among Participants in Mobility-as-a-Service
Journal
Proceedings - 2023 IEEE International Conference on Metaverse Computing, Networking and Applications, MetaCom 2023
ISBN
9798350333336
Date Issued
2023-01-01
Author(s)
Chang, Yu Wei
Abstract
Mobility-as-a-Service (MaaS) is a novel concept integrating multimodal transportation and various service providers (SP) to provide a seamless and efficient transportation experience. However, current MaaS solutions face several challenges, including a mismatch between users and SPs, a lack of cooperation between SPs, users' privacy, and the SPs' business secrets. To address these deficiencies, we present a Federated Dynamic Match-Making Algorithm (FDMMA), which provides multimodal transportation itinerary options with constraint satisfaction, federated multitask learning, and dynamic match-making. Considering SPs' different business goals in addition to users' preferences, FDMMA can encourage SPs to cooperate and compete (co-opetition) in MaaS. Experimental results on a real-world dataset show that the effectiveness of FDMMA can improve the business goals of SPs while maintaining a comparable performance of matching users' selection to other related works.
Subjects
Dynamic Match-Making | Federated Multitask Learning | Mobility-as-a-service | Multimodal transportation
Type
conference paper