Probe-then-Plan: Environment-Aware Planning for Industrial E-commerce Search

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• The proposed Environment-Aware Search Planning (EASP) framework resolves the blindness-latency dilemma in e-commerce search by introducing a Probe-then-Plan mechanism.method 90%
• EASP significantly improves relevant recall and achieves substantial lifts in UCVR and GMV compared to existing methods.result 90%
• The Planner model is trained via Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) alignment with business outcomes like conversion rate.method 85%
• EASP has been successfully deployed in JD.com's AI-Search system.market 95%
• Existing LLM-based paradigms face a fundamental blindness-latency dilemma where query rewriting is agnostic to retrieval capabilities and real-time inventory.limitation 85%
• Deep search agents relying on iterative tool calls and reflection incur seconds of latency incompatible with industrial sub-second budgets.limitation 80%
• The system uses a complexity-aware routing mechanism that selectively activates planning for complex queries to ensure optimal resource allocation.technical 85%
• The method begins with offline data synthesis where a Teacher Agent synthesizes diverse, execution-validated plans by diagnosing the probed environment.method 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card