Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Synthetic Data Powers Product Retrieval for Long-tail Knowledge-Intensive Queries in E-commerce Search

Gui Ling, Weiyuan Li, Yue Jiang, Wenjun Peng, Xingxian Liu, Dongshuai Li, Fuyu Lv, Dan Ou, Haihong Tang

TL;DR

This work proposes an efficient data synthesis framework tailored to retrieval involving long-tail, knowledge-intensive queries, and trains a multi-candidate query rewriting model with multiple reward signals to capture its rewriting capability in well-curated query-product pairs through a powerful offline retrieval pipeline.

Abstract

Product retrieval is the backbone of e-commerce search: for each user query, it identifies a high-recall candidate set from billions of items, laying the foundation for high-quality ranking and user experience. Despite extensive optimization for mainstream queries, existing systems still struggle with long-tail queries, especially knowledge-intensive ones. These queries exhibit diverse linguistic patterns, often lack explicit purchase intent, and require domain-specific knowledge reasoning for accurate interpretation. They also suffer from a shortage of reliable behavioral logs, which makes such queries a persistent challenge for retrieval optimization. To address these issues, we propose an efficient data synthesis framework tailored to retrieval involving long-tail, knowledge-intensive queries. The key idea is to implicitly distill the capabilities of a powerful offline query-rewriting model into an efficient online retrieval system. Leveraging the strong language understanding of LLMs, we train a multi-candidate query rewriting model with multiple reward signals and capture its rewriting capability in well-curated query-product pairs through a powerful offline retrieval pipeline. This design mitigates distributional shift in rewritten queries, which might otherwise limit incremental recall or introduce irrelevant products. Experiments demonstrate that without any additional tricks, simply incorporating this synthetic data into retrieval model training leads to significant improvements. Online Side-By-Side (SBS) human evaluation results indicate a notable enhancement in user search experience.

Synthetic Data Powers Product Retrieval for Long-tail Knowledge-Intensive Queries in E-commerce Search

TL;DR

This work proposes an efficient data synthesis framework tailored to retrieval involving long-tail, knowledge-intensive queries, and trains a multi-candidate query rewriting model with multiple reward signals to capture its rewriting capability in well-curated query-product pairs through a powerful offline retrieval pipeline.

Abstract

Product retrieval is the backbone of e-commerce search: for each user query, it identifies a high-recall candidate set from billions of items, laying the foundation for high-quality ranking and user experience. Despite extensive optimization for mainstream queries, existing systems still struggle with long-tail queries, especially knowledge-intensive ones. These queries exhibit diverse linguistic patterns, often lack explicit purchase intent, and require domain-specific knowledge reasoning for accurate interpretation. They also suffer from a shortage of reliable behavioral logs, which makes such queries a persistent challenge for retrieval optimization. To address these issues, we propose an efficient data synthesis framework tailored to retrieval involving long-tail, knowledge-intensive queries. The key idea is to implicitly distill the capabilities of a powerful offline query-rewriting model into an efficient online retrieval system. Leveraging the strong language understanding of LLMs, we train a multi-candidate query rewriting model with multiple reward signals and capture its rewriting capability in well-curated query-product pairs through a powerful offline retrieval pipeline. This design mitigates distributional shift in rewritten queries, which might otherwise limit incremental recall or introduce irrelevant products. Experiments demonstrate that without any additional tricks, simply incorporating this synthetic data into retrieval model training leads to significant improvements. Online Side-By-Side (SBS) human evaluation results indicate a notable enhancement in user search experience.
Paper Structure (7 sections, 2 equations, 2 figures, 4 tables)

This paper contains 7 sections, 2 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Experiments
  4. Experiment Setup
  5. Offline Evaluation
  6. Online Evaluation
  7. Conclusion

Figures (2)

  • Figure 1: Our proposed data synthesis framework targeting long-tail knowledge-intensive queries. The framework comprises two stages: (1) Multi-Candidate Query Rewriting and (2) Offline Product Retrieval. Query rewriting is the core component of the framework. The retrieval model and the business-specific query-product relevance model are optimized production models aligned with online serving system, while the query–rewrite relevance model and the general-purpose query-product relevance model are open-source LLMs without domain-specific fine-tuning.
  • Figure 2: A multi-reward design for rewriting model optimization. In each iteration, multiple responses (each containing a rewrite list) are sampled and scored on relevance, alignment, and diversity. The three rewards are then weighted and summed to form the final reward for model updates.