{"paper":{"title":"On Integrating Resilience and Human Oversight into LLM-Assisted Modeling Workflows for Digital Twins","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Three design principles using a density-preserving intermediate representation enable resilient LLM-assisted workflows for digital twins.","cross_cats":["cs.AI","cs.SE","cs.SY"],"primary_cat":"eess.SY","authors_text":"Lekshmi P, Neha Karanjkar","submitted_at":"2026-03-26T20:38:03Z","abstract_excerpt":"LLM-assisted modeling holds the potential to rapidly build executable Digital Twins of complex systems from only coarse descriptions and sensor data. However, resilience to LLM hallucination, human oversight, and real-time model adaptability remain challenging and often mutually conflicting requirements. We present three critical design principles for integrating resilience and oversight into such workflows, derived from insights gained through our work on FactoryFlow - an open-source LLM-assisted framework for building simulation-based Digital Twins of manufacturing systems. First, orthogonal"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We present three critical design principles for integrating resilience and oversight into such workflows... A key contribution is detailed characterization of LLM-induced errors across model descriptions of varying detail and complexity, revealing how IR choice critically impacts error rates.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the error characterization and design principles observed in FactoryFlow generalize beyond the specific manufacturing examples and LLM versions tested.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Three design principles for LLM-assisted digital twin modeling emphasize separating structure from parameters, using library components, and a density-preserving Python IR to limit hallucination error growth.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Three design principles using a density-preserving intermediate representation enable resilient LLM-assisted workflows for digital twins.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"85fd763d1ff78d607052e10f7861407703106db47d44c2dcdf8ba458ae425ac6"},"source":{"id":"2603.25898","kind":"arxiv","version":3},"verdict":{"id":"6eed7cbf-2b84-4312-8245-6cfbd49e9f74","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T00:09:25.782911Z","strongest_claim":"We present three critical design principles for integrating resilience and oversight into such workflows... A key contribution is detailed characterization of LLM-induced errors across model descriptions of varying detail and complexity, revealing how IR choice critically impacts error rates.","one_line_summary":"Three design principles for LLM-assisted digital twin modeling emphasize separating structure from parameters, using library components, and a density-preserving Python IR to limit hallucination error growth.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the error characterization and design principles observed in FactoryFlow generalize beyond the specific manufacturing examples and LLM versions tested.","pith_extraction_headline":"Three design principles using a density-preserving intermediate representation enable resilient LLM-assisted workflows for digital twins."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2603.25898/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"}