{"paper":{"title":"CompleteRXN: Toward Completing Open Chemical Reaction Databases","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A new benchmark and constrained model complete missing parts of chemical reactions from open databases at up to 99% accuracy.","cross_cats":["physics.chem-ph"],"primary_cat":"cs.LG","authors_text":"Evgeny Pidko, Gabriel Vogel, Jana M. Weber, Minouk Noordsij","submitted_at":"2026-04-30T20:53:29Z","abstract_excerpt":"Chemical reaction datasets such as USPTO suffer from substantial incompleteness, frequently missing byproducts, co-reactants, and stoichiometric coefficients. This limits their applicability and reliability in downstream applications. Here, we introduce CompleteRXN, a large-scale supervised benchmark for reaction completion under realistic missing-data conditions. We construct a dataset of aligned incomplete and atom-balanced reactions by mapping USPTO records to curated mechanistic reactions. We evaluate representative baselines, including a novel encoder-decoder reaction completion model wit"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"On our CompleteRXN benchmark, the CRB achieves high performance across splits of increasing difficulty, reaching 99.20% equivalence accuracy on the random split and 91.12% on the extreme out-of-distribution split.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The dataset of aligned incomplete and atom-balanced reactions constructed by mapping USPTO records to curated mechanistic reactions accurately captures realistic missing-data conditions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper presents the CompleteRXN benchmark for reaction completion and shows that a novel Constrained Reaction Balancer model reaches 99.20% accuracy on random splits but drops to 91.12% on extreme out-of-distribution cases and further on uncurated data.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A new benchmark and constrained model complete missing parts of chemical reactions from open databases at up to 99% accuracy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"07dadd953d2019d886863ec69e834a4e94aaa4a79e61ae43ff695bb3b0a757e4"},"source":{"id":"2605.00222","kind":"arxiv","version":2},"verdict":{"id":"8d7f6282-82c6-4c44-9c35-720926e345d8","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T20:22:01.714639Z","strongest_claim":"On our CompleteRXN benchmark, the CRB achieves high performance across splits of increasing difficulty, reaching 99.20% equivalence accuracy on the random split and 91.12% on the extreme out-of-distribution split.","one_line_summary":"The paper presents the CompleteRXN benchmark for reaction completion and shows that a novel Constrained Reaction Balancer model reaches 99.20% accuracy on random splits but drops to 91.12% on extreme out-of-distribution cases and further on uncurated data.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The dataset of aligned incomplete and atom-balanced reactions constructed by mapping USPTO records to curated mechanistic reactions accurately captures realistic missing-data conditions.","pith_extraction_headline":"A new benchmark and constrained model complete missing parts of chemical reactions from open databases at up to 99% accuracy."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.00222/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T20:36:25.141035Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:23:15.168047Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9fdee4fdb3dee89547fe11968961ff5ff6295298a73f49f1d8c786220315c5ab"},"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"}