Facebook Marketplace improves diversity with personalized category limits
PCap: Personalized Retrieval-Stage Diversity Capping in Facebook Marketplace
Information Retrieval
Summary
Showing a variety of items helps people enjoy shopping more, but everyone likes different mixes of choices. The authors created a system that learns how much variety each person prefers and limits how many items from each category appear when searching. This system uses math to understand preferences and adjusts limits automatically. Tests on Facebook Marketplace showed this approach makes users more engaged and happier with what they see.
What this means in practice
- •For e-commerce platform engineers: Improve product search diversity by setting personalized category limits during candidate retrieval to enhance user engagement.
- •For online advertising teams: Use personalized diversity limits to better control the variety of ads shown to individual users, balancing relevance and variety.
Authors
Guangchao Yuan, Janis Fuh, Christopher Choate, Xun Tang, Wenqi Zhu, Chengyi Zhang, Pavan Kumar Paalya Chandrashekar, Jiang Han, Jiangyuan Li, Hongyan Wang, Shuting Wang
Abstract
We propose a personalized capping framework (PCap) to improve the diversity in Facebook Marketplace by introducing user-level diversity constraints at the retrieval stage. PCap models individual diversity preferences using Shannon entropy-based scoring, segments users into diversity buckets, and applies personalized category caps during multi-source candidate retrieval. To navigate the high-dimensional parameter space of per-bucket caps, we leverage an automated online optimization method called Parameter Tuning Sequence. Large-scale online experiments demonstrate that PCap significantly improves users' browsing experience shown in engagement metrics. This work provides practical insights into integrating personalized diversity into industrial retrieval systems.