{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ICB6T36ZQETPCKJZPP3WC5SZCD","short_pith_number":"pith:ICB6T36Z","schema_version":"1.0","canonical_sha256":"4083e9efd98126f129397bf761765910f4b7d67c18b24f2795b9bc75c43c93a2","source":{"kind":"arxiv","id":"2505.12837","version":1},"attestation_state":"computed","paper":{"title":"The Hidden Structure -- Improving Legal Document Understanding Through Explicit Text Formatting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alexander Lilienbeck, Christian Braun, Daniel Mentjukov","submitted_at":"2025-05-19T08:25:21Z","abstract_excerpt":"Legal contracts possess an inherent, semantically vital structure (e.g., sections, clauses) that is crucial for human comprehension but whose impact on LLM processing remains under-explored. This paper investigates the effects of explicit input text structure and prompt engineering on the performance of GPT-4o and GPT-4.1 on a legal question-answering task using an excerpt of the CUAD. We compare model exact-match accuracy across various input formats: well-structured plain-text (human-generated from CUAD), plain-text cleaned of line breaks, extracted plain-text from Azure OCR, plain-text extr"},"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":"2505.12837","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-19T08:25:21Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cd6aa180b7792c1004fa93f9693163a10e3dfd66a645f353277fc5b44dcdce84","abstract_canon_sha256":"253829219c75cbcba43f0d9e88ffd9ca87691a6f45c7f5d4e7888ddd86ec4323"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:05:10.441820Z","signature_b64":"JZiSudLgpy7JGTzcxZAv89gUDF9sWd+2C1eVPrCh6oTUea6/5uy29Nyuna8HPRlRpdYJy9dNmAKAQvh57gLHAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4083e9efd98126f129397bf761765910f4b7d67c18b24f2795b9bc75c43c93a2","last_reissued_at":"2026-07-05T11:05:10.441422Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:05:10.441422Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Hidden Structure -- Improving Legal Document Understanding Through Explicit Text Formatting","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alexander Lilienbeck, Christian Braun, Daniel Mentjukov","submitted_at":"2025-05-19T08:25:21Z","abstract_excerpt":"Legal contracts possess an inherent, semantically vital structure (e.g., sections, clauses) that is crucial for human comprehension but whose impact on LLM processing remains under-explored. This paper investigates the effects of explicit input text structure and prompt engineering on the performance of GPT-4o and GPT-4.1 on a legal question-answering task using an excerpt of the CUAD. We compare model exact-match accuracy across various input formats: well-structured plain-text (human-generated from CUAD), plain-text cleaned of line breaks, extracted plain-text from Azure OCR, plain-text extr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.12837","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/2505.12837/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":"2505.12837","created_at":"2026-07-05T11:05:10.441482+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.12837v1","created_at":"2026-07-05T11:05:10.441482+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.12837","created_at":"2026-07-05T11:05:10.441482+00:00"},{"alias_kind":"pith_short_12","alias_value":"ICB6T36ZQETP","created_at":"2026-07-05T11:05:10.441482+00:00"},{"alias_kind":"pith_short_16","alias_value":"ICB6T36ZQETPCKJZ","created_at":"2026-07-05T11:05:10.441482+00:00"},{"alias_kind":"pith_short_8","alias_value":"ICB6T36Z","created_at":"2026-07-05T11:05:10.441482+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07539","citing_title":"Prompt Governance? On Governing Technologies Governed by Natural Language","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23788","citing_title":"MIRAGE: A Micro-Interaction Relational Architecture for Grounded Exploration in Multi-Figure Artworks","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD","json":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD.json","graph_json":"https://pith.science/api/pith-number/ICB6T36ZQETPCKJZPP3WC5SZCD/graph.json","events_json":"https://pith.science/api/pith-number/ICB6T36ZQETPCKJZPP3WC5SZCD/events.json","paper":"https://pith.science/paper/ICB6T36Z"},"agent_actions":{"view_html":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD","download_json":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD.json","view_paper":"https://pith.science/paper/ICB6T36Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.12837&json=true","fetch_graph":"https://pith.science/api/pith-number/ICB6T36ZQETPCKJZPP3WC5SZCD/graph.json","fetch_events":"https://pith.science/api/pith-number/ICB6T36ZQETPCKJZPP3WC5SZCD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD/action/storage_attestation","attest_author":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD/action/author_attestation","sign_citation":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD/action/citation_signature","submit_replication":"https://pith.science/pith/ICB6T36ZQETPCKJZPP3WC5SZCD/action/replication_record"}},"created_at":"2026-07-05T11:05:10.441482+00:00","updated_at":"2026-07-05T11:05:10.441482+00:00"}