Agent4POI achieves 23.2% relative gain over baselines on POI benchmarks

LLM-EDT improves cross-domain sequential recommendation by up to 12.4%

Paper updated on arXiv (v2) for LLM-EDT dual-phase training framework