Mind the Gap: Bridging Behavioral Silos with LLMs in Multi-Vertical Recommendations
For e-commerce platforms with multiple verticals, this work provides a method to leverage LLMs for cross-vertical knowledge transfer, improving recommendation quality in new or sparse verticals.
This paper addresses the cold-start problem in multi-vertical e-commerce platforms by using LLMs to generate features from data-rich verticals (restaurants) to improve recommendations in data-sparse verticals (grocery, retail). The approach achieves significant improvements in personalization and engagement, as demonstrated through offline and online evaluations.
In multi-vertical e-commerce platforms like DoorDash, relatively newer product verticals such as grocery and retail present a significant opportunity for personalization innovation. A key challenge lies in solving the "cold start" problem for users. This paper introduces a novel framework for enhancing recommendation quality by transferring knowledge from data-rich verticals (e.g., restaurants at DoorDash) to data-sparse ones. We leverage Large Language Models (LLMs) to perform generative inference, synthesizing sparse, high-dimensional features that encapsulate latent user affinities. Specifically, we employ a hierarchical Retrieval-Augmented Generation (RAG) pipeline to derive multi-level taxonomic features from user restaurant order histories and search queries. These generated features, encoding both long-term cross-vertical preferences and short-term intent, are integrated into a production Multi-Task Learning (MTL) ranking model. We demonstrate through extensive offline and online evaluation that this approach significantly improves personalization and engagement in emerging business verticals, effectively bridging the behavioral data gap.