{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.23459", "kind": "paper", "url": "https://arxiv.org/abs/2606.23459"}, {"id": "gh:kristenzhang/triggerbench-official", "kind": "repo", "url": "https://github.com/KristenZHANG/TriggerBench-Official"}], "authors": ["Tianhua Zhang", "Xinjiang Wang", "Qianxi Zhang", "Qi Chen", "Kun Li", "Yaoqi Chen", "Dingdong Wang", "Helen Meng", "Yan Lu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.23459", "source_url": "https://arxiv.org/abs/2606.23459"}, "categories": ["General AI", "Language & Knowledge", "cs.CL"], "description": {"en": "TriggerBench is a benchmark for evaluating prospective memory (PM) in LLMs, spanning five dimensions across daily assistant and professional workflow scenarios, with matched retrospective memory (RM) controls, contrastive variants, and overloaded triggers, measuring proactive recall, false-alarm rate, and attentional robustness."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.23459", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.23459", "title": "TriggerBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.23459", "modality": null, "name": "TriggerBench", "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": "unknown", "origin_record_id": "bm_triggerbench_0852166e", "origin_source": "arxiv", "recognition_confidence": "0.95", "record_sha256": "a666d1bc42aa975f056e759786d85445552a5c697053f1a4fab94de85e078a4a", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.23459"}, "publisher": null, "released": "2026-06-22", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.23459", "source_url": "https://arxiv.org/abs/2606.23459"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-23459", "source": "claire_radar", "source_benchmark_id": "2606.23459", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/KristenZHANG/TriggerBench-Official", "githubScope": "benchmark_repo", "githubStars": 2, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2606.23459", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-10-04", "state": "stale"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "60a37abf835e8cac58ae5bad8246d90bf5bc54172cfcc4586079dbbfe3f46c38", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "TriggerBench is a benchmark for evaluating prospective memory (PM) in LLMs, spanning five dimensions across daily assistant and professional workflow scenarios, with matched retrospective memory (RM) controls, contrastive variants, and overloaded triggers, measuring proactive recall, false-alarm rate, and attentional robustness.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "public artifact URL"], "snippet": "We introduce TriggerBench, a comprehensive PM benchmark spanning five dimensions across both daily assistants and professional workflows."}, "familyId": "bmf_85eecdfdb4f5", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/KristenZHANG/TriggerBench-Official", "url": "https://github.com/KristenZHANG/TriggerBench-Official"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/KristenZHANG/TriggerBench-Official", "url": "https://huggingface.co/papers/2606.23459"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/KristenZHANG/TriggerBench-Official.", "url": "https://arxiv.org/abs/2606.23459"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/KristenZHANG/TriggerBench-Official", "url": "https://arxiv.org/abs/2606.23459"}], "scope": "benchmark_repo", "url": "https://github.com/KristenZHANG/TriggerBench-Official", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_triggerbench_0852166e", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/KristenZHANG/TriggerBench-Official", "data": null, "hfPaper": "https://huggingface.co/papers/2606.23459", "paper": "https://arxiv.org/abs/2606.23459", "pdf": "https://arxiv.org/pdf/2606.23459", "project": null, "report": "https://arxiv.org/abs/2606.23459"}, "metrics": [], "motivation": "While Large Language Models (LLMs) are increasingly deployed in long interactions, existing evaluations focus predominantly on retrospective memory (RM) via explicit queries.", "name": "TriggerBench", "oneLine": "TriggerBench is a benchmark for evaluating prospective memory (PM) in LLMs, spanning five dimensions across daily assistant and professional workflow scenarios, with matched retrospective memory (RM) controls, contrastive variants, and overloaded triggers, measuring proactive recall, false-alarm rate, and attentional robustness.", "paperTitle": "TriggerBench: Investigating Prospective Memory for Large Language Models", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.23459", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "KristenZHANG (GitHub)", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://github.com/KristenZHANG/TriggerBench-Official"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 0.95, "relation": "introduces", "releasedAt": "2026-06-22", "source": {"authors": ["Tianhua Zhang", "Xinjiang Wang", "Qianxi Zhang", "Qi Chen", "Kun Li", "Yaoqi Chen", "Dingdong Wang", "Helen Meng", "Yan Lu"], "categories": ["cs.CL"], "id": "2606.23459", "title": "TriggerBench: Investigating Prospective Memory for Large Language Models", "type": "arxiv", "url": "https://arxiv.org/abs/2606.23459"}, "sourceUpdatedAt": "2026-06-22T15:07:36Z", "topics": ["cs.CL"], "venueAttempts": [], "whyItMatters": "Existing LLM evaluations focus on retrospective memory via explicit queries, leaving prospective memory – the ability to spontaneously act on latent constraints – unevaluated. TriggerBench provides a granular measurement of PM capabilities, revealing a precision-recall trade-off, attentional fragility, and a decay with context length that RM does not exhibit, informing deployment decisions for long interactive applications."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
