{"paper":{"title":"A Grammar of Machine Learning Workflows: Rejecting Data Leakage at Call Time","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A grammar of eight typed primitives, a directed acyclic graph, and four hard constraints makes the most damaging forms of data leakage structurally unrepresentable in machine learning workflows.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Simon Roth","submitted_at":"2026-03-11T13:15:33Z","abstract_excerpt":"Data leakage has been identified in 648 published papers across 30 scientific fields. The knowledge to prevent it has existed for over a decade; the problem persists because the tools do not enforce what the textbooks teach. This paper presents a grammar (eight typed primitives connected by a directed acyclic graph with four hard constraints) that makes the most damaging leakage types structurally unrepresentable within the grammar's scope. The core mechanism is a terminal assessment gate: the first call-time-enforced evaluate/assess boundary documented in the peer-reviewed ML methodology lite"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"This paper presents a grammar - eight typed primitives, a directed acyclic graph, and four hard constraints - that makes the most damaging leakage types structurally unrepresentable.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the four hard constraints and the terminal assessment gate are sufficient to block all damaging leakage types while remaining practical for real workflows and enforceable at call time.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A grammar of eight typed primitives, a DAG, and four constraints makes the most damaging forms of data leakage in ML workflows structurally unrepresentable.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A grammar of eight typed primitives, a directed acyclic graph, and four hard constraints makes the most damaging forms of data leakage structurally unrepresentable in machine learning workflows.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"2a2f964aa01480e8387c4bc9efc7cdb2da49b32640b5314caaf9fab0c109ed50"},"source":{"id":"2603.10742","kind":"arxiv","version":4},"verdict":{"id":"66d5934c-ac97-4ea8-ba63-bff7a712dbd9","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T12:33:42.415116Z","strongest_claim":"This paper presents a grammar - eight typed primitives, a directed acyclic graph, and four hard constraints - that makes the most damaging leakage types structurally unrepresentable.","one_line_summary":"A grammar of eight typed primitives, a DAG, and four constraints makes the most damaging forms of data leakage in ML workflows structurally unrepresentable.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the four hard constraints and the terminal assessment gate are sufficient to block all damaging leakage types while remaining practical for real workflows and enforceable at call time.","pith_extraction_headline":"A grammar of eight typed primitives, a directed acyclic graph, and four hard constraints makes the most damaging forms of data leakage structurally unrepresentable in machine learning workflows."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.10742/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":2,"snapshot_sha256":"a1433eb4cf9c0f589123c5c5100b804eab4e49b1d2e194ef855ce2b9ea34226e"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}