{"slug": "from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in", "title": "From Relevant Memories to Sufficient Evidence: Rethinking Memory Utilization in LLM Agents", "summary": "A new arXiv paper (2610.22668v1) introduces QEMem, a framework that reframes LLM agent memory utilization as building query-specific evidence sets rather than retrieving individually relevant memories, targeting what the authors call the relevance-sufficiency gap. QEMem combines query-guided discovery across multiple memory granularities, source-grounded reification into structured ViewCards with explicit evidential roles and source references, and sufficiency-guided refinement to recover missing information, operating on existing memory stores without modifying memory formation. The framework reaches 91.43% LLM-judge accuracy on LoCoMo with Claude Opus 4.5, beating the strongest evaluated baseline, MemR3, by 7.6 percentage points and NEMORI by 10.7 points under a shared memory store.", "body_md": "arXiv:2610.22668v1 Announce Type: new \nAbstract: Long-term memory enables LLM agents to reuse past experiences, yet retrieving relevant memories does not necessarily provide sufficient evidence for answering a query. Essential information may remain fragmented across memory records, obscured by abstraction, or disconnected from the relations required for reasoning. We identify this relevance-sufficiency gap and propose QEMem, a framework that reformulates memory utilization as the construction of query-specific evidence sets, shifting the focus from individual retrieval relevance to collective evidential support. QEMem integrates three complementary mechanisms: query-guided discovery retrieves evidence across multiple memory granularities; source-grounded reification organizes it into structured ViewCards with explicit evidential roles and source references; and sufficiency-guided refinement selectively recovers missing information. Operating on existing memory stores without modifying memory formation, QEMem achieves 91.43% LLM-judge accuracy on LoCoMo with Claude Opus 4.5, outperforming the strongest evaluated baseline, MemR$^3$, by 7.6 percentage points and NEMORI by 10.7 points under a shared memory store. Further analyses demonstrate improved evidence coverage and strong performance on questions requiring cross-session evidence integration.", "url": "https://wpnews.pro/news/from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in", "canonical_source": "https://arxiv.org/abs/2610.22668", "published_at": "2026-10-19 04:00:00+00:00", "updated_at": "2026-10-19 04:18:47.567180+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research", "machine-learning"], "entities": ["QEMem", "LoCoMo", "Claude Opus 4.5", "MemR3", "NEMORI", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in", "markdown": "https://wpnews.pro/news/from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in.md", "text": "https://wpnews.pro/news/from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in.txt", "jsonld": "https://wpnews.pro/news/from-relevant-memories-to-sufficient-evidence-rethinking-memory-utilization-in.jsonld"}}