Search papers, labs, and topics across Lattice.
This paper introduces ICEGR, an Intent-Coherent End-to-End Generative Retrieval Framework designed to enhance e-commerce search by maintaining query-intent consistency throughout the training process. The framework addresses limitations in existing methods, such as static semantic ID construction and inadequate supervision for low-exposure products, by incorporating query-intent signals and augmenting training with synthetic queries. Experimental results demonstrate significant improvements in retrieval performance metrics and real-world business outcomes, including a 3.52% increase in click-through rate and a 15.96% rise in order volume when deployed in Baidu E-commerce Search.
Integrating query intent throughout the generative retrieval pipeline led to a remarkable 21.7% boost in recall and a 3.52% increase in click-through rates in real-world applications.
Generative Retrieval (GR) is promising for e-commerce search, yet existing methods struggle to maintain query-intent consistency throughout the training pipeline. First, semantic ID (SID) construction based on static product information limits the ability of SIDs to encode product-intent associations. Second, although supervised fine-tuning (SFT) learns product-SID mappings across the catalog, low-exposure products still lack real query-intent supervision because query-to-SID training relies solely on online logs, resulting in poor retrieval performance for these products. Third, business-oriented preference optimization may favor popular or high-value products over those that best match the query intent, weakening query-product relevance. To address these issues, we propose ICEGR, an Intent-Coherent End-to-End Generative Retrieval Framework for E-commerce Search that integrates query intent consistently throughout the GR training pipeline. ICEGR comprises three components: (1) Intent-Aware SID Construction incorporates query-intent signals into SID construction, enabling SIDs to capture search intent beyond static product information; (2) Synthetic Query-Enhanced Unified SFT unifies multiple SFT tasks under the query-to-SID objective and augments sparse supervision from online logs with synthetic queries, providing complementary query-intent supervision for low-exposure products; and (3) Relevance-Calibrated Preference Optimization integrates query-product relevance and business signals into a margin-adaptive preference objective, preserving query intent while enabling business preference learning. Offline results show that ICEGR improves Recall@20 by 21.7% and NDCG@20 by 26.6% over the baseline. Deployed as an end-to-end generative retrieval pathway in Baidu E-commerce Search, ICEGR achieves relative improvements of 3.52% in CTR, 15.96% in order volume, and 7.53% in GMV in an A/B test.
