Optimal transport improves multimodal recommendation by modeling diverse user interests
OT-FlowRec: Optimal Transport-Guided Multi-Bridge Flow for Multimodal Sequential Recommendation
Information RetrievalArtificial Intelligence
Summary
Users shopping online often have multiple interests at the same time, which can change as they browse. Existing recommendation systems struggle to understand these mixed and evolving preferences. The authors created OT-FlowRec, a method that models multiple distinct user interests separately using optimal transport math to better recommend products. Their approach keeps these varied interests distinct instead of blending them together, leading to more accurate suggestions. They tested it on real datasets and found it outperformed other top recommendation methods.
What this means in practice
- •For e-commerce platform engineers: Improve recommendation systems by modeling multiple concurrent user interests for better personalized product suggestions.$Commercial implications: Enables e-commerce companies to sell more effectively by offering precise recommendations based on diverse and evolving user preferences.
- •For digital marketing teams: Generate targeted campaigns using more detailed user interest profiles derived from multimodal browsing behavior.
Authors
Miao Sun
Abstract
E-commerce user preferences evolve dynamically within sessions, exhibiting concurrent and multi-centric interest distributions. While existing multimodal sequential recommendation methods leverage global attention to compress historical interactions, they inadequately capture the concurrent nature of multi-interest behaviors and fail to model the underlying multi-centric distribution of user intent. To tackle this issue, we propose OT-FlowRec, an optimal transport-guided multi-bridge flow matching framework that explicitly models multi-interest distributions to facilitate multimodal sequential recommendation. Unlike single-path generative models, OT-FlowRec constructs a mixture of parallel, interest-specific transport flows to model the multi-centric distribution characteristics, where each flow preserves a distinct user interest cluster via an optimal transport plan. Then, a geometric transport mechanism is designed to establish minimum-cost paths from the latent multi-interest distribution to target preference prototypes to provide multi-interest awareness. Moreover, a progressive time-sampling warmup strategy dynamically biases optimization weights toward the final states of the generation process, bridging the gap between continuous global distribution fitting and practical top-$k$ ranking metrics. This approach achieves explicit interest disentanglement, ensuring the preservation of a multi-peaked preference representation instead of collapsing it into a unimodal average. Extensive experiments on three public datasets demonstrate that OT-FlowRec outperforms existing state-of-the-art baselines.