{"evidence_status": "illustrative", "round_id": "001", "policy_id": "P001", "sample_id": "D1", "reference_comparison": "compared", "review_scope": "title", "findings": [{"finding_id": "F01", "kind": "extraction_error", "field": "title", "rule_ids": ["R02"], "evidence": "The full caption runs across the top. Percent of population is printed beside the vertical axis.", "assessment": "The output promotes the axis label to title. The human reference supports the caption interpretation.", "human_attention": false, "uncertainty": null, "attention_reason": "The layout distinguishes the caption and axis label. The council can consider a policy amendment without a reference decision."}]}
{"evidence_status": "illustrative", "round_id": "001", "policy_id": "P001", "sample_id": "D2", "reference_comparison": "compared", "review_scope": "variables and interpretation note", "findings": [{"finding_id": "F02", "kind": "extraction_error", "field": "variables", "rule_ids": ["R03"], "evidence": "Coef and SE are adjacent column headings under One lag and Two lags. Monsoon Onset is a row label.", "assessment": "SE describes the statistic for a concept in the row. The output treats it as a separate variable.", "human_attention": false, "uncertainty": null, "attention_reason": "The column and row headings establish the statistical role of SE. The finding can inform a policy amendment without adjudication."}, {"finding_id": "F03", "kind": "ambiguity", "field": "interpretive_notes", "rule_ids": ["R01", "R04"], "evidence": "The D2 data snapshot has stars and a footer directing the reader to the notes to table 1.", "assessment": "The extraction preserves the same visible cross-reference as the gold annotation. The image does not establish what the thresholds are.", "human_attention": true, "uncertainty": "The image leaves the meaning of the significance markers open. The notes to table 1 would be needed to establish the thresholds.", "attention_reason": "Assigning thresholds would affect interpretation of the estimates. Ask the reviewer whether the visible cross-reference suffices for the metadata."}]}
{"evidence_status": "illustrative", "round_id": "001", "policy_id": "P001", "sample_id": "C1", "reference_comparison": "unavailable", "review_scope": "title, median age unit and year categories", "findings": []}
{"evidence_status": "illustrative", "round_id": "001", "policy_id": "P001", "sample_id": "C2", "reference_comparison": "unavailable", "review_scope": "variables", "findings": [{"finding_id": "F04", "kind": "policy_violation", "field": "variables", "rule_ids": ["R01", "R04"], "evidence": "The bars have numerical labels and the vertical scale has no unit title. The legend shows HIGH and LOW_MIDDLE.", "assessment": "The image does not identify countries as the counted entity or print countries as a unit.", "human_attention": false, "uncertainty": "The counted entity remains unspecified in the data snapshot.", "attention_reason": "P001 allows unsupported values to remain unset. The council can consider this grounding issue without a reference or schema decision."}]}
