Search papers, labs, and topics across Lattice.
This paper introduces UniSGR, a unified framework that enhances semantic ID generation and ranking in recommendation systems by addressing the challenges of fine-grained multi-objective ranking. The framework employs a two-stage training approach, combining a pre-training phase on diverse business scenarios with a scenario-specific alignment phase that optimizes both Value-Aware Parallel Multi-Token Prediction and a unified ranking module. Key results from extensive experiments on a large-scale e-commerce platform show that UniSGR significantly improves ranking efficiency and scalability compared to traditional methods.
UniSGR achieves a breakthrough in recommendation systems by seamlessly integrating semantic ID generation with multi-objective ranking, leading to superior performance in e-commerce applications.
Recommendation systems play a pivotal role in modern e-commerce platforms. While generative retrieval has emerged as a promising paradigm for alleviating the limitations of multi-stage cascade architectures, existing methods still struggle with fine-grained multi-objective ranking. To bridge this gap, we propose UniSGR, a Unified framework for Semantic ID Generation and Ranking. UniSGR adopts a two-stage training paradigm: a multi-scenario pre-training stage that learns from mixed business-scenario data, followed by a scenario-specific alignment stage that jointly optimizes Value-Aware Parallel Multi-Token Prediction (VA-PMTP) and a unified multi-objective ranking module. To better align generation with downstream ranking, we introduce Task-Aware Tokens (TAT) guided by Funnel-Aware Contrastive Learning. Furthermore, we propose Semantic Tree Attention with Reorganized KV cache (STARK), an inference strategy that removes key efficiency bottlenecks in conventional beam search. Extensive offline experiments on a large-scale e-commerce platform demonstrate the effectiveness and scalability of UniSGR.
DAMO