Relevance Evaluation on E-commerce production queries (randomized samples)
4.36Win-Rate GainCase-Driven Multi-Agent Framework
Evaluation Results
| Method | Links | |
|---|---|---|
| Case-Driven Multi-Agent FrameworkPhase=Phase III2026.05 | 4.36 | |
| Case-Driven Multi-Agent FrameworkPhase=Phase II2026.05 | 3.1 | |
| Case-Driven Multi-Agent FrameworkPhase=Phase I2026.05 | 2.71 |