{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.30508", "kind": "paper", "url": "https://arxiv.org/abs/2608.30508"}, {"id": "gh:peijunallin/utilmem", "kind": "repo", "url": "https://github.com/peijunallin/UtilMem"}], "authors": ["Peijun Qing", "Fobo Shi", "Soroush Vosoughi"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.30508", "source_url": "https://arxiv.org/abs/2608.30508"}, "categories": ["General AI", "Language & Knowledge", "Reasoning"], "description": {"en": "Evaluates long-term conversational memory systems on evidence utilization through 1,717 instances across five domains, testing reasoning, implicit retrieval, synthesis, and interference resistance."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.30508", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.30508", "title": "UTILMEM"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.30508", "modality": null, "name": "UTILMEM", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_utilmem_d6cec83c", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "55c77690d18c57cb8fb29318391742a0583770f436a415f440c1b39fa064c61d", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-02T05:23:11.715633Z", "source_url": "https://arxiv.org/abs/2608.30508"}, "publisher": null, "released": "2026-08-31", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.30508", "source_url": "https://arxiv.org/abs/2608.30508"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-30508", "source": "claire_radar", "source_benchmark_id": "2608.30508", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-09-02", "attemptedAt": "2026-09-02T05:12:51.621571Z", "githubRepo": "https://github.com/peijunallin/UtilMem", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.30508", "observedAt": "2026-09-02T05:12:51.621571Z", "signalStatus": {"githubStars": {"observedAt": "2026-09-02T05:12:51.621571Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-09-02T05:12:51.621571Z", "state": "unavailable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "EMNLP acceptance and Hugging Face dataset provide visibility; topic of long-term memory is timely for conversational agents.", "score": 50}, "capabilities": ["Reasoning"], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-02T05:23:11.715633Z", "inputHash": "6ec632b19683a22efe3c42db0f8a488eacfa51242a67ac5a7e1f49453461541f", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "UTILMEM: Benchmarking Evidence Utilization in Long-Term Conversational Memory", "canonicalNameSource": "paper_title", "decisionReason": "Dataset is available on Hugging Face with clear instance counts and tasks; despite pending evaluation code, the frozen data and specified metrics support reusable benchmarking.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-02T05:23:11.715633Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates long-term conversational memory systems on evidence utilization through 1,717 instances across five domains, testing reasoning, implicit retrieval, synthesis, and interference resistance.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "Here, we introduce UtilMem, a diagnostic benchmark comprising 1,717 instances across five domains, designed to evaluate four underexplored aspects of memory utilization: reasoning over dense histories, identifying implicitly relevant memories, synthesizing distributed evidence into summaries, analyses, or plans, and resisting interference from semantically similar distractors."}, "familyId": "bmf_04206a801a80", "firstSeenAt": "2026-09-01", "heat": null, "id": "bm_utilmem_d6cec83c", "indexedAt": "2026-09-01T06:28:29.895112Z", "industrySectors": [], "links": {"code": "https://github.com/peijunallin/UtilMem", "data": null, "paper": "https://arxiv.org/abs/2608.30508", "pdf": "https://arxiv.org/pdf/2608.30508", "project": null, "report": "https://arxiv.org/abs/2608.30508"}, "metrics": [], "motivation": "Long-term memory is increasingly important for conversational agents, yet existing benchmarks primarily measure memory through pointwise factual recall: whether a system can recover isolated facts or event-level details from prior interactions.", "name": "UTILMEM", "oneLine": "Evaluates long-term conversational memory systems on evidence utilization through 1,717 instances across five domains, testing reasoning, implicit retrieval, synthesis, and interference resistance.", "paperTitle": "UTILMEM: Benchmarking Evidence Utilization in Long-Term Conversational Memory", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.30508", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-01T06:35:20.746269Z"}, "publications": [], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 119, "score": 23}, "method": "attention-ranking-v13", "rank": 119, "score": 23}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 312, "score": 23}, "method": "attention-ranking-v13", "rank": 312, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-31", "source": {"authors": ["Peijun Qing", "Fobo Shi", "Soroush Vosoughi"], "categories": ["cs.CL"], "id": "2608.30508", "title": "UTILMEM: Benchmarking Evidence Utilization in Long-Term Conversational Memory", "type": "arxiv", "url": "https://arxiv.org/abs/2608.30508"}, "sourceUpdatedAt": "2026-08-31T09:41:26Z", "topics": ["Reasoning"], "venueAttempts": [], "whyItMatters": "Reveals that conventional factual memory benchmarks do not capture effective integration of distributed evidence, guiding development of memory systems that can use stored information in task-oriented outputs."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
