{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RVLXFYW7PDGAWZ6R7DL33A3UDY","short_pith_number":"pith:RVLXFYW7","schema_version":"1.0","canonical_sha256":"8d5772e2df78cc0b67d1f8d7bd83741e10aab29e477eaf8194bdedec62613702","source":{"kind":"arxiv","id":"2507.15152","version":1},"attestation_state":"computed","paper":{"title":"What Level of Automation is \"Good Enough\"? A Benchmark of Large Language Models for Meta-Analysis Data Extraction","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anuradha Mathrani, Lingbo Li, Teo Susnjak","submitted_at":"2025-07-20T23:09:04Z","abstract_excerpt":"Automating data extraction from full-text randomised controlled trials (RCTs) for meta-analysis remains a significant challenge. This study evaluates the practical performance of three LLMs (Gemini-2.0-flash, Grok-3, GPT-4o-mini) across tasks involving statistical results, risk-of-bias assessments, and study-level characteristics in three medical domains: hypertension, diabetes, and orthopaedics. We tested four distinct prompting strategies (basic prompting, self-reflective prompting, model ensemble, and customised prompts) to determine how to improve extraction quality. All models demonstrate"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.15152","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-20T23:09:04Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"3c1148b53a01b485af5727beab78fd8e2089a72a54322b045c4cec41e7cbc8fa","abstract_canon_sha256":"7a39647715aa4dba4c246607fb794bdeb75d3b1a7c490ed061bc343ae4d22a95"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:04:38.939090Z","signature_b64":"+7jq828yk53FsFQZh/puniwey0iwhwoCN4v0Qd/VGOjfAvGWNI3Eup+3aYbRnzVzpKvqH1dLPRkeqFkLY2PKAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8d5772e2df78cc0b67d1f8d7bd83741e10aab29e477eaf8194bdedec62613702","last_reissued_at":"2026-06-09T01:04:38.938621Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:04:38.938621Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Level of Automation is \"Good Enough\"? A Benchmark of Large Language Models for Meta-Analysis Data Extraction","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Anuradha Mathrani, Lingbo Li, Teo Susnjak","submitted_at":"2025-07-20T23:09:04Z","abstract_excerpt":"Automating data extraction from full-text randomised controlled trials (RCTs) for meta-analysis remains a significant challenge. This study evaluates the practical performance of three LLMs (Gemini-2.0-flash, Grok-3, GPT-4o-mini) across tasks involving statistical results, risk-of-bias assessments, and study-level characteristics in three medical domains: hypertension, diabetes, and orthopaedics. We tested four distinct prompting strategies (basic prompting, self-reflective prompting, model ensemble, and customised prompts) to determine how to improve extraction quality. All models demonstrate"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.15152","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2507.15152/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.15152","created_at":"2026-06-09T01:04:38.938678+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.15152v1","created_at":"2026-06-09T01:04:38.938678+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.15152","created_at":"2026-06-09T01:04:38.938678+00:00"},{"alias_kind":"pith_short_12","alias_value":"RVLXFYW7PDGA","created_at":"2026-06-09T01:04:38.938678+00:00"},{"alias_kind":"pith_short_16","alias_value":"RVLXFYW7PDGAWZ6R","created_at":"2026-06-09T01:04:38.938678+00:00"},{"alias_kind":"pith_short_8","alias_value":"RVLXFYW7","created_at":"2026-06-09T01:04:38.938678+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.00038","citing_title":"Compiling Prompts, Not Crafting Them: A Reproducible Workflow for AI-Assisted Evidence Synthesis","ref_index":4,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY","json":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY.json","graph_json":"https://pith.science/api/pith-number/RVLXFYW7PDGAWZ6R7DL33A3UDY/graph.json","events_json":"https://pith.science/api/pith-number/RVLXFYW7PDGAWZ6R7DL33A3UDY/events.json","paper":"https://pith.science/paper/RVLXFYW7"},"agent_actions":{"view_html":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY","download_json":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY.json","view_paper":"https://pith.science/paper/RVLXFYW7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.15152&json=true","fetch_graph":"https://pith.science/api/pith-number/RVLXFYW7PDGAWZ6R7DL33A3UDY/graph.json","fetch_events":"https://pith.science/api/pith-number/RVLXFYW7PDGAWZ6R7DL33A3UDY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY/action/storage_attestation","attest_author":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY/action/author_attestation","sign_citation":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY/action/citation_signature","submit_replication":"https://pith.science/pith/RVLXFYW7PDGAWZ6R7DL33A3UDY/action/replication_record"}},"created_at":"2026-06-09T01:04:38.938678+00:00","updated_at":"2026-06-09T01:04:38.938678+00:00"}