UniRec improves app recommendations by matching preferences across stages
UniRec: Cross-stage Multi-Task Fusion with Preference Alignment for Cascaded Recommender Systems
Information Retrieval
Summary
Recommender systems often work in steps, where an early step picks items and a later step ranks them. Sometimes these steps don’t agree well, causing good items to get dropped early or rankings to be off. The authors propose uniRec, a new method that trains these steps together, sharing information and aligning their preferences so they work better as a team. This approach helps make better recommendations and increased app usage on a large platform.
What this means in practice
- •For mobile app developers: Enhance recommendation consistency by training pre-ranking and ranking models together to boost user engagement and app usage duration.
- •For e-commerce platform teams: Implement unified fusion of multiple stages in recommendation pipelines to improve item selection and ranking alignment for better customer experiences.
Authors
Lingyuan Kong, Jiaqi Cui, Fanjiao Zeng, Congqi Wang, Yu Li, Yuan Cheng, Jingxin Liu, Xiaoshuang Chen, Kaiqiao Zhan
Abstract
Industrial recommender systems use cascaded stages with different objectives, feature spaces, and latency constraints. Optimizing pre-ranking and ranking separately can create cross-stage inconsistency: upstream models may filter out items preferred by downstream rankers, and independently tuned downstream fusion can offset upstream improvements. Existing multi-task fusion methods focus on multi-objective fusion within the ranking stage, and cross-stage methods typically only add a downstream score factor to upstream ranking. Joint optimization of fusion modules across both stages remains largely unexplored. We propose UniRec, a Unified Cross-stage Recommendation Fusion model. First, the two fusion agents partially share input embeddings and are trained in a single computation graph, so gradients from either stage propagate through the shared representation and influence the other. Second, we introduce a dual-axis preference alignment objective: a vertical cross-stage consistency term transfers downstream pairwise preferences to the upstream fusion score, and a horizontal compact aggregation term reorganizes dozens of pairwise objectives over heterogeneous prior signals into bidirectional preference evidence. Third, we find that unconstrained end-to-end fusion optimization can exploit imbalances in item attribute distributions, over-concentrating on high-reward regions at the cost of other objectives. We therefore add an attribute group-relative regularization that computes advantages within attribute groups and normalizes the policy over the same groups, so uniformly promoting an entire high-reward group yields no optimization gain. Offline, UniRec consistently outperforms single-stage fusion and cross-stage coordination baselines. Online A/B tests show a 0.616\% gain in app usage duration. UniRec is fully deployed on the Kuaishou platform.