AI RESEARCH
Rethinking How to Remember: Beyond Atomic Facts in Lifelong LLM Agent Memory
arXiv CS.CL
•
ArXi:2605.19952v1 Announce Type: new To enable reliable long-term interaction, LLM agents require a memory system that can faithfully, efficiently retrieve, and deeply reason over accumulated dialogue history. Most existing methods adopt an extracted fact based paradigm: handcrafted static prompts compress raw dialogues into atomic facts, which are then d, matched, and injected into downstream reasoning. Nevertheless, such fact-centric designs inevitably discard fine-grained details in original dialogues and fail to deep reasoning over scattered isolated facts.