{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:WWDXT3E5Y3BRGAOOXHHMO7AMHJ","short_pith_number":"pith:WWDXT3E5","schema_version":"1.0","canonical_sha256":"b58779ec9dc6c31301ceb9cec77c0c3a5340b076879e024ff4575a4aaffea7fd","source":{"kind":"arxiv","id":"2310.10627","version":1},"attestation_state":"computed","paper":{"title":"Factored Verification: Detecting and Reducing Hallucination in Summaries of Academic Papers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Andreas Stuhlm\\\"uller, Charlie George","submitted_at":"2023-10-16T17:51:17Z","abstract_excerpt":"Hallucination plagues even frontier LLMs--but how bad is it really for summarizing academic papers? We evaluate Factored Verification, a simple automated method for detecting hallucinations in abstractive summaries. This method sets a new SotA on hallucination detection in the summarization task of the HaluEval benchmark, achieving 76.2% accuracy. We then use this method to estimate how often language models hallucinate when summarizing across multiple academic papers and find 0.62 hallucinations in the average ChatGPT (16k) summary, 0.84 for GPT-4, and 1.55 for Claude 2. We ask models to self"},"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":"2310.10627","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-16T17:51:17Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e83957b0bfe29a847be1f756a71051cee320fb99972fdd8eac184bfdee40bbe9","abstract_canon_sha256":"4999fb6121e7c8119c141aa93a83f7692a67c27784108008e95564b35b8697fb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:01:22.549237Z","signature_b64":"ymt6IPE7KqSpP1vSa0SXAIHkATc1Hw6mG38ICjrr1nlIIkPilpCBQBFyOomYpppTIlj6rgXQs5LX2q2YUt0PAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b58779ec9dc6c31301ceb9cec77c0c3a5340b076879e024ff4575a4aaffea7fd","last_reissued_at":"2026-07-05T07:01:22.548751Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:01:22.548751Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Factored Verification: Detecting and Reducing Hallucination in Summaries of Academic Papers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Andreas Stuhlm\\\"uller, Charlie George","submitted_at":"2023-10-16T17:51:17Z","abstract_excerpt":"Hallucination plagues even frontier LLMs--but how bad is it really for summarizing academic papers? We evaluate Factored Verification, a simple automated method for detecting hallucinations in abstractive summaries. This method sets a new SotA on hallucination detection in the summarization task of the HaluEval benchmark, achieving 76.2% accuracy. We then use this method to estimate how often language models hallucinate when summarizing across multiple academic papers and find 0.62 hallucinations in the average ChatGPT (16k) summary, 0.84 for GPT-4, and 1.55 for Claude 2. We ask models to self"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.10627","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/2310.10627/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":"2310.10627","created_at":"2026-07-05T07:01:22.548811+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.10627v1","created_at":"2026-07-05T07:01:22.548811+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.10627","created_at":"2026-07-05T07:01:22.548811+00:00"},{"alias_kind":"pith_short_12","alias_value":"WWDXT3E5Y3BR","created_at":"2026-07-05T07:01:22.548811+00:00"},{"alias_kind":"pith_short_16","alias_value":"WWDXT3E5Y3BRGAOO","created_at":"2026-07-05T07:01:22.548811+00:00"},{"alias_kind":"pith_short_8","alias_value":"WWDXT3E5","created_at":"2026-07-05T07:01:22.548811+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.05965","citing_title":"OpenFActScore: Open-Source Atomic Evaluation of Factuality in Text Generation","ref_index":7,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ","json":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ.json","graph_json":"https://pith.science/api/pith-number/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/graph.json","events_json":"https://pith.science/api/pith-number/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/events.json","paper":"https://pith.science/paper/WWDXT3E5"},"agent_actions":{"view_html":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ","download_json":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ.json","view_paper":"https://pith.science/paper/WWDXT3E5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.10627&json=true","fetch_graph":"https://pith.science/api/pith-number/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/graph.json","fetch_events":"https://pith.science/api/pith-number/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/action/storage_attestation","attest_author":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/action/author_attestation","sign_citation":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/action/citation_signature","submit_replication":"https://pith.science/pith/WWDXT3E5Y3BRGAOOXHHMO7AMHJ/action/replication_record"}},"created_at":"2026-07-05T07:01:22.548811+00:00","updated_at":"2026-07-05T07:01:22.548811+00:00"}