{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.29477", "kind": "paper", "url": "https://arxiv.org/abs/2608.29477"}, {"id": "gh:luoojason/muddle", "kind": "repo", "url": "https://github.com/luoojason/muddle"}], "authors": ["Jason Luo", "Saibilila Abudukelimu", "Judy Song", "Andrew Feng", "Shivank Garg", "Vasu Sharma", "Kevin Zhu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.29477", "source_url": "https://arxiv.org/abs/2608.29477"}, "categories": ["General AI", "Language & Knowledge", "cs.CL"], "description": {"en": "MUDDLE separates the effects of distractor topical similarity and added context length on document question answering, using 270 questions across five controlled conditions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.29477", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.29477", "title": "MUDDLE"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.29477", "modality": null, "name": "MUDDLE", "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_muddle_d01dfe23", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "59b4b728dc44758f01d55c368e60e1c844a0bdb2676680a20f752840a8c52840", "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/2608.29477"}, "publisher": null, "released": "2026-08-30", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.29477", "source_url": "https://arxiv.org/abs/2608.29477"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-29477", "source": "claire_radar", "source_benchmark_id": "2608.29477", "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/luoojason/muddle", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.29477", "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": "Document QA robustness under distractors and long contexts is an active research area, and the availability of code and data supports rapid community uptake.", "score": 70}, "capabilities": [], "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": "7dda7c4fab79f903b14dc4596aa099b7518a71e3f96dfc16222dc0bcd8768d92", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "# MUDDLE Data-construction and evaluation code for **MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects**", "canonicalNameSource": "official_readme", "decisionReason": "The benchmark provides a detailed construction pipeline, released data and evaluation code, and a consistent scoring protocol with deterministic distractor selection.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-02T05:23:11.715633Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "MUDDLE separates the effects of distractor topical similarity and added context length on document question answering, using 270 questions across five controlled conditions.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We present MUDDLE, a controlled benchmark that separates them."}, "familyId": "bmf_598ecf758980", "firstSeenAt": "2026-09-01", "heat": null, "id": "bm_muddle_d01dfe23", "indexedAt": "2026-09-02T05:21:06.631990Z", "industrySectors": [], "links": {"code": "https://github.com/luoojason/muddle", "data": null, "paper": "https://arxiv.org/abs/2608.29477", "pdf": "https://arxiv.org/pdf/2608.29477", "project": null, "report": "https://arxiv.org/abs/2608.29477"}, "metrics": [], "motivation": "Document question-answering systems increasingly answer questions over collections of retrieved documents rather than one clean source, so robustness to distracting context matters as much as reading ability.", "name": "MUDDLE", "oneLine": "MUDDLE separates the effects of distractor topical similarity and added context length on document question answering, using 270 questions across five controlled conditions.", "paperTitle": "MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects", "primaryDomain": "General AI", "publication": {"evidence": "12 pages. Accepted to the Context Beyond the Window (CBW) workshop at COLM 2026 (non-archival). Code and data: https://github.com/luoojason/muddle", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2608.29477", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "Context Beyond the Window (CBW) workshop at COLM 2026 (non-archival)", "verifiedAt": "2026-09-02T05:21:06.631990Z"}, "publications": [], "publishers": [{"name": "Jason Luo", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/luoojason/muddle"}], "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": 123, "score": 23}, "method": "attention-ranking-v13", "rank": 123, "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": 316, "score": 23}, "method": "attention-ranking-v13", "rank": 316, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-30", "source": {"authors": ["Jason Luo", "Saibilila Abudukelimu", "Judy Song", "Andrew Feng", "Shivank Garg", "Vasu Sharma", "Kevin Zhu"], "categories": ["cs.CL", "cs.AI"], "id": "2608.29477", "title": "MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects", "type": "arxiv", "url": "https://arxiv.org/abs/2608.29477"}, "sourceUpdatedAt": "2026-08-30T00:10:40Z", "topics": ["cs.CL"], "venueAttempts": [{"decisionRaw": "12 pages. Accepted to the Context Beyond the Window (CBW) workshop at COLM 2026 (non-archival). Code and data: https://github.com/luoojason/muddle", "evidence": [{"level": "author-claim", "observedAt": "2026-09-02T05:21:06.631990Z", "rawValue": "12 pages. Accepted to the Context Beyond the Window (CBW) workshop at COLM 2026 (non-archival). Code and data: https://github.com/luoojason/muddle", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2608.29477"}], "reviewStatus": "accepted", "venueName": "Context Beyond the Window (CBW) workshop at COLM 2026 (non-archival)"}], "whyItMatters": "It provides a controlled setup to diagnose whether document QA failures stem from topic-related distractors or context length, enabling targeted model improvements."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
