Paper Number
ICIS2025-2291
Paper Type
Complete
Abstract
Users of online dating platforms increasingly experience burnout. This is partly driven by core principles of recommender systems, which reinforce sorting patterns based on explicit attributes such as social status, age, or geographic distance. Accordingly, we propose a serendipity-oriented recommender system by (1) relaxing constraints on explicit attributes while maintaining relevance through personalized preference learning, and (2) leveraging an attention network to extract intrinsic compatibility from profile Q&A texts. In a randomized field experiment on a leading online dating platform, users in the treatment group exhibited a 5.2% relative decrease in profile views and a 35.8% relative reduction in match requests, but achieved a 29.4% relative increase in match probability. Additionally, they showed a 38.5% relative increase in positive emotion scores. They also significantly expanded their consideration sets and formed reciprocal matches more efficiently. Our findings offer causal evidence that incorporating serendipity into reciprocal recommendation can mitigate user burnout.
Recommended Citation
He, Yumei; Guan, Yue; Huang, Ao; and Huang, Nina, "How to Design Serendipity for Burnout Mitigation? A Serendipity-Oriented Recommender System and Field Experiment" (2025). ICIS 2025 Proceedings. 12.
https://aisel.aisnet.org/icis2025/general_topic/general_topic/12
How to Design Serendipity for Burnout Mitigation? A Serendipity-Oriented Recommender System and Field Experiment
Users of online dating platforms increasingly experience burnout. This is partly driven by core principles of recommender systems, which reinforce sorting patterns based on explicit attributes such as social status, age, or geographic distance. Accordingly, we propose a serendipity-oriented recommender system by (1) relaxing constraints on explicit attributes while maintaining relevance through personalized preference learning, and (2) leveraging an attention network to extract intrinsic compatibility from profile Q&A texts. In a randomized field experiment on a leading online dating platform, users in the treatment group exhibited a 5.2% relative decrease in profile views and a 35.8% relative reduction in match requests, but achieved a 29.4% relative increase in match probability. Additionally, they showed a 38.5% relative increase in positive emotion scores. They also significantly expanded their consideration sets and formed reciprocal matches more efficiently. Our findings offer causal evidence that incorporating serendipity into reciprocal recommendation can mitigate user burnout.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
02-GeneralTopics