{ "schema": "lri.security-claim-traceability-matrix.v1", "claim_count": 6, "edge_count": 20, "claims": [ { "claim_id": "rc_01_molot_is_direct_mainline", "claim": "MOLOT 已可作为安全专题的直接论文主线，但它证明的是 Python/JavaScript 包 SAST 静态恶意代码检测，不证明 FCD C/C++ 表。", "level": "L2_direct_public_evidence", "promotion_status": "promoted", "decision_use": "顶部结论应从“候选模型”升级为“直接论文证据主线”。", "evidence_refs": [ "molot_system_card" ], "source_urls": [ "https://arxiv.org/abs/2606.07792" ], "evidence_nodes": [ { "card_id": "molot_system_card", "title": "MOLOT System Card: Malicious Operational Logic Observation Transformer", "source_url": "https://arxiv.org/abs/2606.07792", "quality": { "evidence_id": "molot_system_card", "authority": "peer_reviewed", "recency": "current", "directness": "direct", "replicability": "public_url", "method_quality": "high", "resource_specificity": "core_model_paper", "risk_of_overclaim": "medium", "allowed_use": "current_claim" }, "direct_evidence_level": "L2_direct_public_evidence", "direct_scope": "直接支撑 MOLOT 作为 SAST 静态恶意代码检测系统、静态调用图行为序列、BERT/Transformer 分类、SHAP 源码位置映射、Python/JavaScript 包评估和 Open Malicious-Code Bench。", "markdown_card": "sources/resource_cards/01_molot_system_card.md" } ], "linked_blockers": [ "block_fcd_pdg_table_upgrade", "block_open_malicious_code_bench_artifact_upgrade", "block_agent_gate_performance_upgrade" ], "blocked_extrapolation": [ "FCD C/C++ performance table", "LRI Agent Gate runtime/FPR", "independent Open Malicious-Code Bench artifact URL" ], "weakest_policy": "blocked_or_boundary_claim", "upgrade_condition": "Resolve linked blockers before any stronger wording is allowed." }, { "claim_id": "rc_02_behavior_chain_becomes_detection_principle", "claim": "行为链检测已经从方法背景上移为安全专题的核心原则：单个 API/关键词不足以判断恶意目的，必须看 activity chain。", "level": "L2_multi_paper_supported_principle", "promotion_status": "promoted", "decision_use": "LRI gate taxonomy 应以 activity_chain 为一等字段，而不是 isolated keyword/API。", "evidence_refs": [ "molot_system_card", "cerebro_behavior_sequence", "npm_malicious_detection_benchmark" ], "source_urls": [ "https://arxiv.org/abs/2606.07792", "https://arxiv.org/abs/2309.02637", "https://arxiv.org/abs/2603.27549" ], "evidence_nodes": [ { "card_id": "molot_system_card", "title": "MOLOT System Card: Malicious Operational Logic Observation Transformer", "source_url": "https://arxiv.org/abs/2606.07792", "quality": { "evidence_id": "molot_system_card", "authority": "peer_reviewed", "recency": "current", "directness": "direct", "replicability": "public_url", "method_quality": "high", "resource_specificity": "core_model_paper", "risk_of_overclaim": "medium", "allowed_use": "current_claim" }, "direct_evidence_level": "L2_direct_public_evidence", "direct_scope": "直接支撑 MOLOT 作为 SAST 静态恶意代码检测系统、静态调用图行为序列、BERT/Transformer 分类、SHAP 源码位置映射、Python/JavaScript 包评估和 Open Malicious-Code Bench。", "markdown_card": "sources/resource_cards/01_molot_system_card.md" }, { "card_id": "cerebro_behavior_sequence", "title": "Malicious Package Detection in NPM and PyPI using a Single Model of Malicious Behavior Sequence", "source_url": "https://arxiv.org/abs/2309.02637", "quality": { "evidence_id": "cerebro_behavior_sequence", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "method_predecessor", "risk_of_overclaim": "medium", "allowed_use": "background" }, "direct_evidence_level": "L1_public_method_background", "direct_scope": "直接支撑 NPM/PyPI 恶意包可以通过高层恶意行为抽象、行为序列和 BERT 统一建模。", "markdown_card": "sources/resource_cards/02_cerebro_behavior_sequence.md" }, { "card_id": "npm_malicious_detection_benchmark", "title": "Understanding NPM Malicious Package Detection: A Benchmark-Driven Empirical Analysis", "source_url": "https://arxiv.org/abs/2603.27549", "quality": { "evidence_id": "npm_malicious_detection_benchmark", "authority": "peer_reviewed", "recency": "current", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "benchmark_paper", "risk_of_overclaim": "medium", "allowed_use": "background" }, "direct_evidence_level": "L2_direct_benchmark_evidence", "direct_scope": "直接支撑 NPM 恶意包 benchmark、行为类别、规避技术、多工具对比和行为链优先判断。", "markdown_card": "sources/resource_cards/04_npm_malicious_detection_benchmark.md" } ], "linked_blockers": [], "blocked_extrapolation": [ "binary-only detection", "exact precision without benchmark table extraction" ], "weakest_policy": "public_evidence_supported", "upgrade_condition": "No upgrade needed for current wording." }, { "claim_id": "rc_03_false_positive_budget_is_required", "claim": "安全门禁不能只追求召回率；误报预算和 block/ask_human 分流必须前置。", "level": "L2_direct_operational_constraint", "promotion_status": "promoted", "decision_use": "Agent Gate MVP 应默认区分 block、warn、ask_human，而非二元恶意/良性。", "evidence_refs": [ "python_malware_benchmark", "npm_malicious_detection_benchmark", "molot_system_card" ], "source_urls": [ "https://arxiv.org/abs/2209.13288", "https://arxiv.org/abs/2603.27549", "https://arxiv.org/abs/2606.07792" ], "evidence_nodes": [ { "card_id": "python_malware_benchmark", "title": "A Benchmark Comparison of Python Malware Detection Approaches", "source_url": "https://arxiv.org/abs/2209.13288", "quality": { "evidence_id": "python_malware_benchmark", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "benchmark_paper", "risk_of_overclaim": "medium", "allowed_use": "background" }, "direct_evidence_level": "L2_direct_baseline_evidence", "direct_scope": "直接支撑 PyPI 恶意包检测 benchmark、Bandit4Mal/OSSGadget/PyPI checks 对比、以及仓库 moderation 对误报率极其敏感的约束。", "markdown_card": "sources/resource_cards/03_python_malware_benchmark_bandit4mal_ossgadget.md" }, { "card_id": "npm_malicious_detection_benchmark", "title": "Understanding NPM Malicious Package Detection: A Benchmark-Driven Empirical Analysis", "source_url": "https://arxiv.org/abs/2603.27549", "quality": { "evidence_id": "npm_malicious_detection_benchmark", "authority": "peer_reviewed", "recency": "current", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "benchmark_paper", "risk_of_overclaim": "medium", "allowed_use": "background" }, "direct_evidence_level": "L2_direct_benchmark_evidence", "direct_scope": "直接支撑 NPM 恶意包 benchmark、行为类别、规避技术、多工具对比和行为链优先判断。", "markdown_card": "sources/resource_cards/04_npm_malicious_detection_benchmark.md" }, { "card_id": "molot_system_card", "title": "MOLOT System Card: Malicious Operational Logic Observation Transformer", "source_url": "https://arxiv.org/abs/2606.07792", "quality": { "evidence_id": "molot_system_card", "authority": "peer_reviewed", "recency": "current", "directness": "direct", "replicability": "public_url", "method_quality": "high", "resource_specificity": "core_model_paper", "risk_of_overclaim": "medium", "allowed_use": "current_claim" }, "direct_evidence_level": "L2_direct_public_evidence", "direct_scope": "直接支撑 MOLOT 作为 SAST 静态恶意代码检测系统、静态调用图行为序列、BERT/Transformer 分类、SHAP 源码位置映射、Python/JavaScript 包评估和 Open Malicious-Code Bench。", "markdown_card": "sources/resource_cards/01_molot_system_card.md" } ], "linked_blockers": [ "block_agent_gate_performance_upgrade" ], "blocked_extrapolation": [ "claiming LRI measured FPR before logs exist" ], "weakest_policy": "blocked_or_boundary_claim", "upgrade_condition": "Resolve linked blockers before any stronger wording is allowed." }, { "claim_id": "rc_04_explainability_is_review_evidence", "claim": "SHAP/解释不应停留在可视化，而应上移为 reviewer evidence：source_locations、activity_chain、reviewer_action 必须绑定。", "level": "L2_direct_plus_foundation_evidence", "promotion_status": "promoted", "decision_use": "reviewer_evidence_schema 应保持为工程化主产物。", "evidence_refs": [ "molot_system_card", "shap_unified_approach" ], "source_urls": [ "https://arxiv.org/abs/2606.07792", "https://arxiv.org/abs/1705.07874" ], "evidence_nodes": [ { "card_id": "molot_system_card", "title": "MOLOT System Card: Malicious Operational Logic Observation Transformer", "source_url": "https://arxiv.org/abs/2606.07792", "quality": { "evidence_id": "molot_system_card", "authority": "peer_reviewed", "recency": "current", "directness": "direct", "replicability": "public_url", "method_quality": "high", "resource_specificity": "core_model_paper", "risk_of_overclaim": "medium", "allowed_use": "current_claim" }, "direct_evidence_level": "L2_direct_public_evidence", "direct_scope": "直接支撑 MOLOT 作为 SAST 静态恶意代码检测系统、静态调用图行为序列、BERT/Transformer 分类、SHAP 源码位置映射、Python/JavaScript 包评估和 Open Malicious-Code Bench。", "markdown_card": "sources/resource_cards/01_molot_system_card.md" }, { "card_id": "shap_unified_approach", "title": "A Unified Approach to Interpreting Model Predictions", "source_url": "https://arxiv.org/abs/1705.07874", "quality": { "evidence_id": "shap_unified_approach", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "explainability_paper", "risk_of_overclaim": "high", "allowed_use": "background" }, "direct_evidence_level": "L1_public_explainability_foundation", "direct_scope": "直接支撑 SHAP 作为 feature attribution / additive explanation 方法，用于解释复杂模型预测。", "markdown_card": "sources/resource_cards/11_shap_unified_approach.md" } ], "linked_blockers": [], "blocked_extrapolation": [ "SHAP alone as causal proof", "SHAP alone deciding block" ], "weakest_policy": "background_with_overclaim_risk", "upgrade_condition": "No upgrade needed for current wording." }, { "claim_id": "rc_05_graph_pdg_is_secondary_chain_not_molot_metric", "claim": "PDG/Graph 证据可作为源码/片段相似检测的第二检测面，但不能上移为 MOLOT 性能或 FCD 表直接证据。", "level": "L1_background_evidence_with_boundary", "promotion_status": "boundary_or_candidate", "decision_use": "工作台应保留 PDG/FCD 线，但所有性能数字必须阻断升级。", "evidence_refs": [ "pdg_holmes_graph_siamese", "source_code_is_graph", "fcd_pdg_performance_unresolved" ], "source_urls": [ "https://arxiv.org/abs/2011.11228", "https://arxiv.org/abs/2312.16488", "user_provided_json" ], "evidence_nodes": [ { "card_id": "pdg_holmes_graph_siamese", "title": "Modeling Functional Similarity in Source Code with Graph-Based Siamese Networks", "source_url": "https://arxiv.org/abs/2011.11228", "quality": { "evidence_id": "pdg_holmes_graph_siamese", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "graph_clone_detection_paper", "risk_of_overclaim": "high", "allowed_use": "background" }, "direct_evidence_level": "L1_public_graph_clone_background", "direct_scope": "直接支撑程序依赖图和图神经网络用于语义代码 clone detection。", "markdown_card": "sources/resource_cards/09_pdg_holmes_graph_siamese_clone_detection.md" }, { "card_id": "source_code_is_graph", "title": "Source Code is a Graph, Not a Sequence: A Cross-Lingual Perspective on Code Clone Detection", "source_url": "https://arxiv.org/abs/2312.16488", "quality": { "evidence_id": "source_code_is_graph", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "graph_clone_detection_paper", "risk_of_overclaim": "high", "allowed_use": "background" }, "direct_evidence_level": "L1_public_graph_clone_background", "direct_scope": "直接支撑源码图表示在 code clone detection，尤其跨语言 clone 上，相比序列模型的优势。", "markdown_card": "sources/resource_cards/10_source_code_is_graph_clone_detection.md" }, { "card_id": "fcd_pdg_performance_unresolved", "title": "FCD / PDG C/C++ Performance Table from User JSON", "source_url": "user_provided_json", "quality": { "evidence_id": "fcd_pdg_performance_unresolved", "authority": "user_input", "recency": "unverified", "directness": "unresolved", "replicability": "not_reproducible", "method_quality": "not_applicable", "resource_specificity": "pending_verification_card", "risk_of_overclaim": "high", "allowed_use": "candidate_only" }, "direct_evidence_level": "L0_user_input_only", "direct_scope": "无直接公网论文证据；只保留为用户 JSON 中的待核验 C/C++ FCD/PDG 性能表线索。", "markdown_card": "sources/resource_cards/12_fcd_pdg_performance_unresolved.md" } ], "linked_blockers": [ "block_fcd_pdg_table_upgrade" ], "blocked_extrapolation": [ "verified C/C++ precision/recall/RMSE/FCD speed", "equating graph clone papers with malicious-code detection metrics" ], "weakest_policy": "blocked_or_boundary_claim", "upgrade_condition": "Resolve linked blockers before any stronger wording is allowed." }, { "claim_id": "rc_06_agent_gate_is_design_candidate_not_verified_result", "claim": "LRI Agent Gate 已具备合同、schema 和 MVP 用例设计，但还不能上移为已验证性能结论。", "level": "L0_to_L1_design_candidate", "promotion_status": "boundary_or_candidate", "decision_use": "顶部结论应写成“工程设计候选”，下一步是跑 MVP audit log。", "evidence_refs": [ "molot_system_card", "python_malware_benchmark" ], "source_urls": [ "https://arxiv.org/abs/2606.07792", "https://arxiv.org/abs/2209.13288" ], "evidence_nodes": [ { "card_id": "molot_system_card", "title": "MOLOT System Card: Malicious Operational Logic Observation Transformer", "source_url": "https://arxiv.org/abs/2606.07792", "quality": { "evidence_id": "molot_system_card", "authority": "peer_reviewed", "recency": "current", "directness": "direct", "replicability": "public_url", "method_quality": "high", "resource_specificity": "core_model_paper", "risk_of_overclaim": "medium", "allowed_use": "current_claim" }, "direct_evidence_level": "L2_direct_public_evidence", "direct_scope": "直接支撑 MOLOT 作为 SAST 静态恶意代码检测系统、静态调用图行为序列、BERT/Transformer 分类、SHAP 源码位置映射、Python/JavaScript 包评估和 Open Malicious-Code Bench。", "markdown_card": "sources/resource_cards/01_molot_system_card.md" }, { "card_id": "python_malware_benchmark", "title": "A Benchmark Comparison of Python Malware Detection Approaches", "source_url": "https://arxiv.org/abs/2209.13288", "quality": { "evidence_id": "python_malware_benchmark", "authority": "peer_reviewed", "recency": "aging", "directness": "background", "replicability": "public_url", "method_quality": "high", "resource_specificity": "benchmark_paper", "risk_of_overclaim": "medium", "allowed_use": "background" }, "direct_evidence_level": "L2_direct_baseline_evidence", "direct_scope": "直接支撑 PyPI 恶意包检测 benchmark、Bandit4Mal/OSSGadget/PyPI checks 对比、以及仓库 moderation 对误报率极其敏感的约束。", "markdown_card": "sources/resource_cards/03_python_malware_benchmark_bandit4mal_ossgadget.md" } ], "linked_blockers": [ "block_agent_gate_performance_upgrade" ], "blocked_extrapolation": [ "verified Agent Gate precision", "verified latency/memory", "verified false-positive budget" ], "weakest_policy": "blocked_or_boundary_claim", "upgrade_condition": "Resolve linked blockers before any stronger wording is allowed." } ], "edges": [ { "from": "rc_01_molot_is_direct_mainline", "to": "molot_system_card", "relation": "supported_by" }, { "from": "rc_01_molot_is_direct_mainline", "to": "block_fcd_pdg_table_upgrade", "relation": "blocked_by" }, { "from": "rc_01_molot_is_direct_mainline", "to": "block_open_malicious_code_bench_artifact_upgrade", "relation": "blocked_by" }, { "from": "rc_01_molot_is_direct_mainline", "to": "block_agent_gate_performance_upgrade", "relation": "blocked_by" }, { "from": "rc_02_behavior_chain_becomes_detection_principle", "to": "molot_system_card", "relation": "supported_by" }, { "from": "rc_02_behavior_chain_becomes_detection_principle", "to": "cerebro_behavior_sequence", "relation": "supported_by" }, { "from": "rc_02_behavior_chain_becomes_detection_principle", "to": "npm_malicious_detection_benchmark", "relation": "supported_by" }, { "from": "rc_03_false_positive_budget_is_required", "to": "python_malware_benchmark", "relation": "supported_by" }, { "from": "rc_03_false_positive_budget_is_required", "to": "npm_malicious_detection_benchmark", "relation": "supported_by" }, { "from": "rc_03_false_positive_budget_is_required", "to": "molot_system_card", "relation": "supported_by" }, { "from": "rc_03_false_positive_budget_is_required", "to": "block_agent_gate_performance_upgrade", "relation": "blocked_by" }, { "from": "rc_04_explainability_is_review_evidence", "to": "molot_system_card", "relation": "supported_by" }, { "from": "rc_04_explainability_is_review_evidence", "to": "shap_unified_approach", "relation": "supported_by" }, { "from": "rc_05_graph_pdg_is_secondary_chain_not_molot_metric", "to": "pdg_holmes_graph_siamese", "relation": "supported_by" }, { "from": "rc_05_graph_pdg_is_secondary_chain_not_molot_metric", "to": "source_code_is_graph", "relation": "supported_by" }, { "from": "rc_05_graph_pdg_is_secondary_chain_not_molot_metric", "to": "fcd_pdg_performance_unresolved", "relation": "supported_by" }, { "from": "rc_05_graph_pdg_is_secondary_chain_not_molot_metric", "to": "block_fcd_pdg_table_upgrade", "relation": "blocked_by" }, { "from": "rc_06_agent_gate_is_design_candidate_not_verified_result", "to": "molot_system_card", "relation": "supported_by" }, { "from": "rc_06_agent_gate_is_design_candidate_not_verified_result", "to": "python_malware_benchmark", "relation": "supported_by" }, { "from": "rc_06_agent_gate_is_design_candidate_not_verified_result", "to": "block_agent_gate_performance_upgrade", "relation": "blocked_by" } ], "blocker_ids": [ "block_fcd_pdg_table_upgrade", "block_open_malicious_code_bench_artifact_upgrade", "block_agent_gate_performance_upgrade" ], "rule": "Every top conclusion must trace to resource cards, source URLs, evidence-quality fields, and any blocker that limits stronger wording." }