{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:75QLMDK3XVTRJNKULP5LL6B2XY","short_pith_number":"pith:75QLMDK3","schema_version":"1.0","canonical_sha256":"ff60b60d5bbd6714b5545bfab5f83abe083cd9a56e6cd00331dbbbc528ac6edf","source":{"kind":"arxiv","id":"2502.16143","version":1},"attestation_state":"computed","paper":{"title":"The Law of Knowledge Overshadowing: Towards Understanding, Predicting, and Preventing LLM Hallucination","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Cheng Qian, ChengXiang Zhai, Chi Han, Heng Ji, Jiateng Liu, Kathleen McKeown, Manling Li, Pengfei Yu, Sha Li, Yi R. Fung, Yuji Zhang","submitted_at":"2025-02-22T08:36:06Z","abstract_excerpt":"Hallucination is a persistent challenge in large language models (LLMs), where even with rigorous quality control, models often generate distorted facts. This paradox, in which error generation continues despite high-quality training data, calls for a deeper understanding of the underlying LLM mechanisms. To address it, we propose a novel concept: knowledge overshadowing, where model's dominant knowledge can obscure less prominent knowledge during text generation, causing the model to fabricate inaccurate details. Building on this idea, we introduce a novel framework to quantify factual halluc"},"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":"2502.16143","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-02-22T08:36:06Z","cross_cats_sorted":[],"title_canon_sha256":"7fabf85410c18957306b24f489a529b3d602486864d14dfc6fbbb85524937cb3","abstract_canon_sha256":"438c6de20aad7a22e8fce2d68e0b030e2123c5b36a9e727ec926b5ba3eea1ff2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:18:21.997840Z","signature_b64":"YzV5Zy/whvdrC5v5zRaMBQKqQdxCQZOGGcluLwP8pGggDLDfmYiySjDh3Jr85Ml0KtEiz2LgXOeeRDVTnTswAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff60b60d5bbd6714b5545bfab5f83abe083cd9a56e6cd00331dbbbc528ac6edf","last_reissued_at":"2026-07-05T10:18:21.997281Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:18:21.997281Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Law of Knowledge Overshadowing: Towards Understanding, Predicting, and Preventing LLM Hallucination","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Cheng Qian, ChengXiang Zhai, Chi Han, Heng Ji, Jiateng Liu, Kathleen McKeown, Manling Li, Pengfei Yu, Sha Li, Yi R. Fung, Yuji Zhang","submitted_at":"2025-02-22T08:36:06Z","abstract_excerpt":"Hallucination is a persistent challenge in large language models (LLMs), where even with rigorous quality control, models often generate distorted facts. This paradox, in which error generation continues despite high-quality training data, calls for a deeper understanding of the underlying LLM mechanisms. To address it, we propose a novel concept: knowledge overshadowing, where model's dominant knowledge can obscure less prominent knowledge during text generation, causing the model to fabricate inaccurate details. Building on this idea, we introduce a novel framework to quantify factual halluc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.16143","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/2502.16143/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":"2502.16143","created_at":"2026-07-05T10:18:21.997350+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.16143v1","created_at":"2026-07-05T10:18:21.997350+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.16143","created_at":"2026-07-05T10:18:21.997350+00:00"},{"alias_kind":"pith_short_12","alias_value":"75QLMDK3XVTR","created_at":"2026-07-05T10:18:21.997350+00:00"},{"alias_kind":"pith_short_16","alias_value":"75QLMDK3XVTRJNKU","created_at":"2026-07-05T10:18:21.997350+00:00"},{"alias_kind":"pith_short_8","alias_value":"75QLMDK3","created_at":"2026-07-05T10:18:21.997350+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.18732","citing_title":"Predictable Confabulations: Factual Recall by LLMs Scales with Model Size and Topic Frequency","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12384","citing_title":"Scalable Token-Level Hallucination Detection in Large Language Models","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12312","citing_title":"CompliBench: Benchmarking LLM Judges for Compliance Violation Detection in Dialogue Systems","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY","json":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY.json","graph_json":"https://pith.science/api/pith-number/75QLMDK3XVTRJNKULP5LL6B2XY/graph.json","events_json":"https://pith.science/api/pith-number/75QLMDK3XVTRJNKULP5LL6B2XY/events.json","paper":"https://pith.science/paper/75QLMDK3"},"agent_actions":{"view_html":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY","download_json":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY.json","view_paper":"https://pith.science/paper/75QLMDK3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.16143&json=true","fetch_graph":"https://pith.science/api/pith-number/75QLMDK3XVTRJNKULP5LL6B2XY/graph.json","fetch_events":"https://pith.science/api/pith-number/75QLMDK3XVTRJNKULP5LL6B2XY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY/action/storage_attestation","attest_author":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY/action/author_attestation","sign_citation":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY/action/citation_signature","submit_replication":"https://pith.science/pith/75QLMDK3XVTRJNKULP5LL6B2XY/action/replication_record"}},"created_at":"2026-07-05T10:18:21.997350+00:00","updated_at":"2026-07-05T10:18:21.997350+00:00"}