{"paper":{"title":"A Deterministic Agentic Workflow for HS Tariff Classification: Multi-Dimensional Rule Reasoning with Interpretable Decisions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A deterministic six-stage agentic workflow for HS tariff classification reaches 64.2% top-1 accuracy at six digits on HSCodeComp and flags possible inconsistencies in some ground-truth labels.","cross_cats":["cs.IR"],"primary_cat":"cs.AI","authors_text":"Dongjiang Zhuang, Jing Cao, Junhe Wu, Kai Chen, Qu Zhou, Yu Zhang, Zheng Huang","submitted_at":"2026-05-14T14:04:46Z","abstract_excerpt":"Harmonized System (HS) tariff classification is a high-stakes, expert-level task in which a free-form product description must be mapped to a specific six- or eight-digit code under the General Interpretive Rules (GIR), section notes, chapter notes, and Explanatory Notes. The difficulty lies not in knowledge volume but in *multi-dimensional rule reasoning*: a correct classification must satisfy competing priority rules along several axes simultaneously, including material, form, function, essential character, the part-versus-whole boundary, and specific listing versus residual headings. End-to"},"claims":{"count":3,"items":[{"kind":"strongest_claim","text":"Evaluated on HSCodeComp at the six-digit level, the workflow reaches 75.0% top-1 and 91.5% top-3 at four digits, and 64.2% top-1 and 78.3% top-3 at six digits with Qwen3.6-plus; an open-weight Qwen3.6-27B-FP8 backbone achieves 84.2% four-digit and 77.4% six-digit top-1 agreement.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a fixed six-stage decomposition of multi-dimensional HS rule reasoning, supported by offline knowledge engineering, will correctly resolve all priority conflicts among material, function, essential character, and part-versus-whole axes without missing interactions that require dynamic reordering.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A deterministic six-stage agentic workflow for HS tariff classification reaches 64.2% top-1 accuracy at six digits on HSCodeComp and flags possible inconsistencies in some ground-truth labels.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"}],"snapshot_sha256":"df916f009e99e56abcd5690bfc1d48b997a914f282a1cfbc0a18c13eb8c241ff"},"source":{"id":"2605.14857","kind":"arxiv","version":1},"verdict":{"id":"9c7ef451-7333-4fc4-bff6-68195fdd95f2","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T03:18:43.811371Z","strongest_claim":"Evaluated on HSCodeComp at the six-digit level, the workflow reaches 75.0% top-1 and 91.5% top-3 at four digits, and 64.2% top-1 and 78.3% top-3 at six digits with Qwen3.6-plus; an open-weight Qwen3.6-27B-FP8 backbone achieves 84.2% four-digit and 77.4% six-digit top-1 agreement.","one_line_summary":"A deterministic six-stage agentic workflow for HS tariff classification reaches 64.2% top-1 accuracy at six digits on HSCodeComp and flags possible inconsistencies in some ground-truth labels.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a fixed six-stage decomposition of multi-dimensional HS rule reasoning, supported by offline knowledge engineering, will correctly resolve all priority conflicts among material, function, essential character, and part-versus-whole axes without missing interactions that require dynamic reordering.","pith_extraction_headline":""},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":1,"snapshot_sha256":"9b675e3c5c29d64067208b39e0842b8bde3ab38b754676b06332bee8e56d9617"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}