SIA: A Synthesize-Inject-Align Framework for Knowledge-Grounded and Secure E-commerce Search LLMs with Industrial Deployment

TL;DR

This paper introduces SIA, a comprehensive framework that synthesizes knowledge, injects domain expertise, and aligns models to enhance the security and knowledgeability of e-commerce search LLMs, validated through large-scale industrial deployment.

Contribution

The paper presents a novel SIA framework combining knowledge synthesis, parameter-efficient pre-training, and dual-path alignment for secure, knowledge-grounded e-commerce search LLMs.

Findings

Significant improvements in key business metrics at JD.com

Effective knowledge injection with Depth Up-Scaling strategy

Enhanced security robustness against jailbreak attacks

Abstract

Large language models offer transformative potential for e-commerce search by enabling intent-aware recommendations. However, their industrial deployment is hindered by two critical challenges: (1) knowledge hallucination due to insufficient encoding of dynamic, fine-grained product knowledge, and (2) security vulnerabilities under jailbreak attacks that threaten compliance. To address these issues, we propose SIA--a Synthesize-Inject-Align framework for building knowledgeable and secure e-commerce search LLMs. Our approach first synthesizes high-quality natural language corpus by combining structured knowledge graphs with unstructured behavioral logs, augmented with reasoning chains and safety-aware data. We then introduce a parameter-efficient pre-training strategy based on Depth Up-Scaling to inject domain knowledge while preserving general capabilities. Finally, a dual-path alignment method via multi-task instruction tuning and adversarial training strengthens both task performance and safety robustness. The framework has been deployed at JD.com, China's largest self-operated e-commerce platform, where A/B tests across five core search scenarios demonstrate significant improvements in key business metrics, validating its industrial effectiveness and scalability.