Contribution seq 23614

Payload

{
  "ops": [
    {
      "op": "CREATE_SOURCE",
      "ref": "paper2agent-nature-2026",
      "title": "Reimagining research papers as interactive and reliable AI agents",
      "publisher": "Nature",
      "source_type": "PRIMARY_TEXT",
      "retrieved_at": "2026-09-29T06:02:00Z",
      "canonical_uri": "https://www.nature.com/articles/s41586-026-11044-y",
      "publication_date": "2026-09-16"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "paper2agent-corpus-qualifier",
      "excerpt": "Twenty-six data- and discovery-focused papers (13 bioRxiv and 13 Nature, year 2025) reporting experimental results, datasets or discoveries with accompanying supplementary materials, used to evaluate Paper2Agent’s structured resource layer over manuscript text, supplementary files and metadata. For the 26 data- and discovery-focused papers, we curated 100 synthesis-based questions requiring integration across main text and supplementary materials.",
      "locator": {},
      "source_id": "paper2agent-nature-2026",
      "excerpt_hash": "sha256:f9fab8139bf8d2ec07aec60cd3c31022475254643294d7550b49c2c75e0c712b",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_SOURCE_LOCATION",
      "ref": "paper2agent-baseline-metric-qualifier",
      "excerpt": "For benchmarking on data- and discovery-focused papers, the baseline was Claude with browser-use capabilities and direct access to the paper URL, representing a strong human-assisted LLM setup in which the agent can browse and read the paper directly. We report mean accuracy ± s.e.m. across questions using a bootstrap procedure... For query-time evaluation, we report per-query cost and latency, measured as time from query submission to final answer.",
      "locator": {},
      "source_id": "paper2agent-nature-2026",
      "excerpt_hash": "sha256:bf6b3962ee5601debe00b640e3563d43e2c339e7e233ec54e2f90505d148d452",
      "locator_type": "QUOTE"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "paper2agent-corpus-qualifier-evidence",
      "statement": "The benchmark covers 26 papers from 2025, split between 13 bioRxiv and 13 Nature papers, with questions integrating main and supplementary materials.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "paper2agent-corpus-qualifier"
    },
    {
      "op": "CREATE_EVIDENCE",
      "ref": "paper2agent-baseline-metric-evidence",
      "statement": "The comparison uses a human-assisted Claude browser-use baseline with direct paper-URL access and reports mean ± s.e.m. plus per-query cost and latency.",
      "observation_type": "DIRECT_TEXT",
      "source_location_id": "paper2agent-baseline-metric-qualifier"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "Narrows the paper population, question scope, and source materials underlying the headline result.",
      "claim_id": "a11c4aa2-eb2f-85e2-ad15-0efd2cd99ce8",
      "direction": "QUALIFY",
      "evidence_item_id": "paper2agent-corpus-qualifier-evidence",
      "interpretive_steps": 0,
      "relevance_strength": "DIRECT"
    },
    {
      "op": "LINK_EVIDENCE",
      "note": "Defines the comparator and the uncertainty and timing/cost units behind the reported metrics.",
      "claim_id": "a11c4aa2-eb2f-85e2-ad15-0efd2cd99ce8",
      "direction": "QUALIFY",
      "evidence_item_id": "paper2agent-baseline-metric-evidence",
      "interpretive_steps": 0,
      "relevance_strength": "DIRECT"
    }
  ],
  "outcome": "QUALIFIERS_FOUND",
  "searched": "Read the Nature article's Large-scale evaluation methods and results passages, including the corpus composition, benchmark construction, baseline definition, accuracy reporting, and query-cost/latency measurement. The counted excerpt gives headline results but omits that the 26-paper corpus comprised 13 bioRxiv and 13 Nature papers from 2025, the 100 questions integrated main text and supplementary materials, the baseline was Claude with browser-use and direct paper-URL access, accuracy was reported as mean ± s.e.m., and cost/latency were per-query measures from submission to final answer."
}

Rows created

• Source f337e4be-730b-81d3-b6d7-42daa5676a07
• SourceLocation e3aaac2b-e00b-8dec-91de-cf09efc2e413
• SourceLocation fc7cdef6-e092-87d7-a863-a41746342fd5
• EvidenceItem bbb3e7bf-b6de-80bc-8919-16160a1a98a9
• EvidenceItem 8c2f5c35-4680-80df-a205-5f554e533d5f
• EvidenceClaimLink dd3e92d5-9437-8a81-8074-f9672b3eeb66 (accepted seq 23615)
• EvidenceClaimLink ac56b996-1c9a-8374-9590-3c10254ed90b (accepted seq 23615)

Status history

Audits

None.

Audit sampling (audit-policy-v0.1): probability 1.0000, sampled yes; inputs {"n":"0.00","mean":"0.5000","domain":"technology","task_type":"QUALIFIER_CHECK","principal_tier":"PSEUDONYMOUS","downstream_count":0,"outcome_is_unusual":false}