{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:QA27J7LX6ZUTKDDLHL7NWJR6HF","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"f75b5fdba43fa6f27656b5148d3ab22de49f099e0985577c26dac837994d783f","cross_cats_sorted":["cs.AI","cs.DL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-03-20T17:11:58Z","title_canon_sha256":"0df2f9ec91101d2ccbd07c10f43a29340eb5d541d5211aaa35e1aa2e7e0f256f"},"schema_version":"1.0","source":{"id":"2603.22327","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.22327","created_at":"2026-06-08T00:03:39Z"},{"alias_kind":"arxiv_version","alias_value":"2603.22327v2","created_at":"2026-06-08T00:03:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.22327","created_at":"2026-06-08T00:03:39Z"},{"alias_kind":"pith_short_12","alias_value":"QA27J7LX6ZUT","created_at":"2026-06-08T00:03:39Z"},{"alias_kind":"pith_short_16","alias_value":"QA27J7LX6ZUTKDDL","created_at":"2026-06-08T00:03:39Z"},{"alias_kind":"pith_short_8","alias_value":"QA27J7LX","created_at":"2026-06-08T00:03:39Z"}],"graph_snapshots":[{"event_id":"sha256:c02ca464555ddd9551881b93c0705d99b26c6148879689e83aa986f7db145260","target":"graph","created_at":"2026-06-08T00:03:39Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2603.22327/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Systematic literature reviews (SLRs) are a demanding and high-stakes form of scientific knowledge synthesis that remains underspecified as an evaluation setting for large language models (LLMs). We introduce AgentSLR, a large-scale evaluation harness comprising an SLR automation workflow and an expert annotated dataset covering 16,248 articles, designed to test LLM capabilities across the stages of SLRs in epidemiology. Reference annotations were derived from peer-reviewed studies on WHO priority pathogens and produced by domain experts. The harness evaluates each review stage as a separate un","authors_text":"Adam Mahdi, Anne Cori, Christian Morgenstern, Elizaveta Semenova, Jakob Foerster, Lingyi Yang, {\\L}ukasz Borchmann, Philip H. Torr, Piotr B{\\l}aszczyk, Ruth McCabe, Ryan Othniel Kearns, Sangeeta Bhatia, Scott A. Hale, Shreyansh Padarha, Thomas Rawson, Tristan Naidoo","cross_cats":["cs.AI","cs.DL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-03-20T17:11:58Z","title":"Evaluating AI-based Scientific Knowledge Synthesis with Epidemiological Systematic Reviews"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.22327","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:488e27f803d7acae1a3195b7de8dcf12502089feafb2d29008d767d2f2e780bc","target":"record","created_at":"2026-06-08T00:03:39Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"f75b5fdba43fa6f27656b5148d3ab22de49f099e0985577c26dac837994d783f","cross_cats_sorted":["cs.AI","cs.DL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IR","submitted_at":"2026-03-20T17:11:58Z","title_canon_sha256":"0df2f9ec91101d2ccbd07c10f43a29340eb5d541d5211aaa35e1aa2e7e0f256f"},"schema_version":"1.0","source":{"id":"2603.22327","kind":"arxiv","version":2}},"canonical_sha256":"8035f4fd77f669350c6b3afedb263e39550f65a1472bf26cd07c30bc959234d5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8035f4fd77f669350c6b3afedb263e39550f65a1472bf26cd07c30bc959234d5","first_computed_at":"2026-06-08T00:03:39.274624Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-08T00:03:39.274624Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"73QElPySgM4EpzXyrXR7FDqsQNwXdMjyYMZkIjO+kMQbuRAbyhFpllqebgkfpwFEzyW1yqS8skFZWp5HxvmrCQ==","signature_status":"signed_v1","signed_at":"2026-06-08T00:03:39.275734Z","signed_message":"canonical_sha256_bytes"},"source_id":"2603.22327","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:488e27f803d7acae1a3195b7de8dcf12502089feafb2d29008d767d2f2e780bc","sha256:c02ca464555ddd9551881b93c0705d99b26c6148879689e83aa986f7db145260"],"state_sha256":"1d5e9d2f9b1d315ee30688f9f1ec6fad4b78ca4d9985eeb8cd6fe73e286ff402"}