{"paper":{"title":"SecGoal: A Benchmark for Extracting Formalizable Security Goals from Protocol Documents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Instruction tuning on SecGoal lets 7B and 9B models extract security goals from protocol documents at over 80% F1 while larger general models fall below 15% precision.","cross_cats":[],"primary_cat":"cs.CR","authors_text":"Bo Jia, Dawei Huang, Haonan Feng, Hui Li, Jingjing Guan, Xiangdong Li, Yueshuang Jiao","submitted_at":"2026-04-30T08:50:03Z","abstract_excerpt":"Formal verification provides rigorous guarantees for cryptographic security, yet extracting formalizable security goals from natural-language protocol documents remains largely manual. We introduce SecGoal, a dedicated expert-annotated dataset and benchmark for extracting formalizable security goal statements from protocol documents, covering 15 widely deployed protocols, together with AIFG, a schema- and flow-conditioned framework for structured formal security property generation. Our evaluation shows that frontier and large LLMs achieve high property recall but low extraction precision beca"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"instruction tuning on SecGoal enables compact models with 7B/9B parameters to achieve F1-scores above 80%, substantially outperforming larger general-purpose models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The expert-annotated security goals in the SecGoal benchmark are accurate, unbiased, and sufficient to train models that generalize to new protocol documents.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"The paper presents SecGoal, the first expert-annotated benchmark for security goal extraction from protocol documents, and demonstrates that fine-tuned 7B/9B parameter models achieve over 80% F1 score, outperforming larger general LLMs.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Instruction tuning on SecGoal lets 7B and 9B models extract security goals from protocol documents at over 80% F1 while larger general models fall below 15% precision.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b0ed20d17053656fd1d59c739184d666e6b4e2c00b42150120523bd9ae6838aa"},"source":{"id":"2604.27601","kind":"arxiv","version":2},"verdict":{"id":"694042aa-0560-4470-ac65-4b76bebe720c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-07T07:21:23.527521Z","strongest_claim":"instruction tuning on SecGoal enables compact models with 7B/9B parameters to achieve F1-scores above 80%, substantially outperforming larger general-purpose models.","one_line_summary":"The paper presents SecGoal, the first expert-annotated benchmark for security goal extraction from protocol documents, and demonstrates that fine-tuned 7B/9B parameter models achieve over 80% F1 score, outperforming larger general LLMs.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The expert-annotated security goals in the SecGoal benchmark are accurate, unbiased, and sufficient to train models that generalize to new protocol documents.","pith_extraction_headline":"Instruction tuning on SecGoal lets 7B and 9B models extract security goals from protocol documents at over 80% F1 while larger general models fall below 15% precision."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.27601/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T21:43:41.392900Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T19:04:00.337887Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"6e079c321a28a812d1624f4d253d9235e6943c257661e6cfa8b1393691da140b"},"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"}