Paper Number
ICIS2025-2513
Paper Type
Complete
Abstract
Teams are often more effective at tackling complex tasks than individuals and many online platforms encourage collaboration. However, participants face many challenges in forming teams, including high search costs and uncertainty about candidates' capabilities and willingness to collaborate. On the other hand, platforms can recommend viable collaborators to requesters by leveraging historical data on participants. When many participants request collaboration recommendations, processing requests arriving close together in a batch can benefit the platform. Doing so without considering the impact on their stakeholders could hurt some more than others – for example, some candidates may be recommended to multiple requesters while others are overlooked. Enforcing candidate-side fairness by limiting how often they are recommended in a given time frame can adversely affect requesters. We propose a team recommender system that stays fair to requesters by minimizing the maximum harm to any requester. Extensive experiments demonstrate the effectiveness of our approach.
Recommended Citation
Liu, Yihong; Menon, Syam; and Sarkar, Sumit, "Batching and Fairness When Recommending Multiple Teams in Online Platforms" (2025). ICIS 2025 Proceedings. 17.
https://aisel.aisnet.org/icis2025/da_bus/da_bus/17
Batching and Fairness When Recommending Multiple Teams in Online Platforms
Teams are often more effective at tackling complex tasks than individuals and many online platforms encourage collaboration. However, participants face many challenges in forming teams, including high search costs and uncertainty about candidates' capabilities and willingness to collaborate. On the other hand, platforms can recommend viable collaborators to requesters by leveraging historical data on participants. When many participants request collaboration recommendations, processing requests arriving close together in a batch can benefit the platform. Doing so without considering the impact on their stakeholders could hurt some more than others – for example, some candidates may be recommended to multiple requesters while others are overlooked. Enforcing candidate-side fairness by limiting how often they are recommended in a given time frame can adversely affect requesters. We propose a team recommender system that stays fair to requesters by minimizing the maximum harm to any requester. Extensive experiments demonstrate the effectiveness of our approach.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
07-DataAnalytics