{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.17391", "kind": "paper", "url": "https://arxiv.org/abs/2606.17391"}], "authors": ["Logan Mann", "Abdur Rahman", "Mohammad Saifullah", "Taaha Kazi", "Vasu Sharma"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.17391", "source_url": "https://arxiv.org/abs/2606.17391"}, "categories": ["General AI", "Speech & Audio", "cs.CL"], "description": {"en": "NarrativeWorldBench evaluates long-horizon narrative generation in audio drama across nine structural metrics and four Indic languages, across horizons from 10 to 200 episodes. Scoring is based on plot-beat F1 and other metrics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.17391", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.17391", "title": "NarrativeWorldBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["University of California, Santa Barbara"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.17391", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.17391", "source_url": "https://openalex.org/W7165043224"}, "key": "claire-radar:2606.17391", "modality": null, "name": "NarrativeWorldBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "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_narrativeworldbench_d47453bd", "origin_source": "arxiv", "recognition_confidence": "0.85", "record_sha256": "a7c4dc7a6e75584e5c188e2e615dedc56a04c5d3ac3a069285bc70d6f4e3bcad", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.17391"}, "publisher": null, "released": "2026-06-16", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.17391", "source_url": "https://arxiv.org/abs/2606.17391"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-17391", "source": "claire_radar", "source_benchmark_id": "2606.17391", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Speech & Audio", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2606.17391", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-21T04:30:40.934319Z", "inputHash": "6996f6ece7ba8c94674684b0437947afc552b16fa5701464ca2ce45ce5b07017", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "NarrativeWorldBench evaluates long-horizon narrative generation in audio drama across nine structural metrics and four Indic languages, across horizons from 10 to 200 episodes. Scoring is based on plot-beat F1 and other metrics.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence"], "snippet": "We introduce NarrativeWorldBench, an open benchmark of nine narrative-structure metrics evaluated across horizons h in {10, 20, 50, 100, 200}, with cross-lingual evaluation across four Indic languages (Hindi, Tamil, Telugu, Marathi)."}, "familyId": "bmf_402c51ce9e0f", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_narrativeworldbench_d47453bd", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.17391", "paper": "https://arxiv.org/abs/2606.17391", "pdf": "https://arxiv.org/pdf/2606.17391", "project": null, "report": "https://arxiv.org/abs/2606.17391"}, "metrics": [], "motivation": "Long-form serialized audio drama, with arcs that run for 200 to 800 episodes, is a major creative medium and a setting where frontier large language models (LLMs) fail.", "name": "NarrativeWorldBench", "oneLine": "NarrativeWorldBench evaluates long-horizon narrative generation in audio drama across nine structural metrics and four Indic languages, across horizons from 10 to 200 episodes. Scoring is based on plot-beat F1 and other metrics.", "paperTitle": "NarrativeWorldBench: A Frontier-Saturated Benchmark and a Latent World Model for Long-Horizon Co-Creative Audio Drama", "primaryDomain": "General AI", "publication": {"evidence": "10 pages. Accepted to the ICML 2026 Workshops on High-dimensional Learning Dynamics (HiLD) and Culture x AI", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2606.17391", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "ICML 2026 Workshops on High-dimensional Learning Dynamics (HiLD) and Culture x AI", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "arXiv", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.17391"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 0.85, "relation": "introduces", "releasedAt": "2026-06-16", "source": {"authors": ["Logan Mann", "Abdur Rahman", "Mohammad Saifullah", "Taaha Kazi", "Vasu Sharma"], "categories": ["cs.CL", "cs.AI", "cs.LG"], "id": "2606.17391", "title": "NarrativeWorldBench: A Frontier-Saturated Benchmark and a Latent World Model for Long-Horizon Co-Creative Audio Drama", "type": "arxiv", "url": "https://arxiv.org/abs/2606.17391"}, "sourceUpdatedAt": "2026-06-16T01:04:55Z", "topics": ["cs.CL"], "venueAttempts": [{"decisionRaw": "10 pages. Accepted to the ICML 2026 Workshops on High-dimensional Learning Dynamics (HiLD) and Culture x AI", "evidence": [{"level": "author-claim", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "10 pages. Accepted to the ICML 2026 Workshops on High-dimensional Learning Dynamics (HiLD) and Culture x AI", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2606.17391"}], "reviewStatus": "accepted", "venueName": "ICML 2026 Workshops on High-dimensional Learning Dynamics (HiLD) and Culture x AI"}], "whyItMatters": "Current LLMs degrade on long-horizon narrative coherence. This benchmark provides metrics for evaluating long-form structured content generation and cross-lingual capabilities."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
