Search papers, labs, and topics across Lattice.
This paper introduces Environment-Aware Search Planning (EASP) to address the blindness-latency dilemma in LLM-based e-commerce search, where query rewriting ignores retrieval capabilities and deep search agents are too slow. EASP uses a "Probe-then-Plan" mechanism, where a retrieval probe exposes the retrieval snapshot to the planner, enabling grounded search plans. Experiments on JD.com show that EASP improves relevant recall and achieves substantial lifts in UCVR and GMV, leading to successful deployment in their AI-Search system.
LLMs can plan effective e-commerce searches within strict latency budgets by first probing the retrieval environment to ground their reasoning.
Modern e-commerce search is evolving to resolve complex user intents. While Large Language Models (LLMs) offer strong reasoning, existing LLM-based paradigms face a fundamental blindness-latency dilemma: query rewriting is agnostic to retrieval capabilities and real-time inventory, yielding invalid plans; conversely, deep search agents rely on iterative tool calls and reflection, incurring seconds of latency incompatible with industrial sub-second budgets. To resolve this conflict, we propose Environment-Aware Search Planning (EASP), reformulating search planning as a dynamic reasoning process grounded in environmental reality. EASP introduces a Probe-then-Plan mechanism: a lightweight Retrieval Probe exposes the retrieval snapshot, enabling the Planner to diagnose execution gaps and generate grounded search plans. The methodology comprises three stages: (1) Offline Data Synthesis: A Teacher Agent synthesizes diverse, execution-validated plans by diagnosing the probed environment. (2) Planner Training and Alignment: The Planner is initialized via Supervised Fine-Tuning (SFT) to internalize diagnostic capabilities, then aligned with business outcomes (conversion rate) via Reinforcement Learning (RL). (3) Adaptive Online Serving: A complexity-aware routing mechanism selectively activates planning for complex queries, ensuring optimal resource allocation. Extensive offline evaluations and online A/B testing on JD.com demonstrate that EASP significantly improves relevant recall and achieves substantial lifts in UCVR and GMV. EASP has been successfully deployed in JD.com's AI-Search system.
