eduevidence 6.2.0 → 6.3.0

package/domains/education/copy/section_titles.json

@@ -0,0 +1,186 @@
+{
+  "domain": "education",
+  "purpose": "Section titles/leads + default full-report chapter plan (education voice).",
+  "section_titles": {
+    "zh": {
+      "01": "01 执行决策",
+      "02": "02 结果证据概览",
+      "03": "03 证据矩阵",
+      "04": "04 证据裁决",
+      "05": "05 方法学审计",
+      "06": "06 冲突分析",
+      "07": "07 主张-证据追溯",
+      "08": "08 适用性",
+      "09": "09 教学干预",
+      "10": "10 评价方案",
+      "11": "11 基准测试",
+      "12": "12 来源与溯源"
+    },
+    "en": {
+      "01": "01 Executive Decision",
+      "02": "02 Outcome Evidence Overview",
+      "03": "03 Evidence Matrix",
+      "04": "04 Evidence Tribunal",
+      "05": "05 Methodology Audit",
+      "06": "06 Conflict Analysis",
+      "07": "07 Claim-Evidence Trace",
+      "08": "08 Applicability",
+      "09": "09 Teaching Intervention",
+      "10": "10 Evaluation Plan",
+      "11": "11 Benchmark",
+      "12": "12 Sources & Provenance"
+    }
+  },
+  "section_leads": {
+    "zh": {
+      "01": "本节先给结论：最终怎么裁决、置信度多高、靠哪几条证据。",
+      "02": "一图看清：哪些学习结果有支持证据、哪些被反驳。",
+      "03": "每条证据来自哪项研究、测了什么、方向与质量如何；可筛选、可搜索。",
+      "04": "证据允许主张什么、不允许主张什么；缺失的关键证据是什么。",
+      "05": "研究质量可靠吗？哪些方法学问题让结论打折。",
+      "06": "不同研究为何结论不同；分歧出在哪一环。",
+      "07": "从结论到证据到原始来源，每一步都可追查。",
+      "08": "结论适用于谁、什么课程与结果、需要什么条件。",
+      "09": "试点怎么分阶段放开 AI 规则；什么情况必须叫停。",
+      "10": "如何验证效果：指标、对照、成功阈值。",
+      "11": "EduEvidence 自身基准表现：引用精度与成本。",
+      "12": "每篇文献是谁、出自哪里、如何获取。"
+    },
+    "en": {
+      "01": "The verdict first: what we decide, at what confidence, on which evidence.",
+      "02": "At a glance: which learning outcomes have supporting evidence, which are contradicted.",
+      "03": "Where each piece of evidence comes from, what it measures, its direction and quality — filterable and searchable.",
+      "04": "What the evidence lets us claim, what it does not, and what is still missing.",
+      "05": "How reliable are these studies, and which methodological concerns discount the conclusions.",
+      "06": "Why studies disagree — and where exactly they diverge.",
+      "07": "Every step from conclusion to evidence to source stays traceable.",
+      "08": "Who the conclusion applies to, for which course and outcomes, under what conditions.",
+      "09": "How AI usage rules phase in during a pilot, and when we must stop.",
+      "10": "How we verify real effects: metrics, comparison, success threshold.",
+      "11": "How EduEvidence itself performs: citation precision and cost.",
+      "12": "Who wrote each cited study, where it came from, how it was fetched."
+    }
+  },
+  "default_chapters": [
+    {
+      "key": "decision",
+      "title_zh": "结论、裁决与研究边界",
+      "title_en": "Decision, Adjudication & Research Boundary",
+      "lead_zh": "先明确最终裁决与研究边界，再解释为什么。",
+      "lead_en": "State the final adjudication and research boundary before explaining why.",
+      "modules": [
+        "decision",
+        "scope"
+      ]
+    },
+    {
+      "key": "evidence",
+      "title_zh": "关键证据与结果分离",
+      "title_en": "Key Evidence & Outcome Separation",
+      "lead_zh": "把任务表现、真实学习、保持与风险放在同一证据地图中，但不混为一谈。",
+      "lead_en": "Place task performance, actual learning, retention and risk on one evidence map without conflating them.",
+      "modules": [
+        "retrieval",
+        "outcomes",
+        "evidence"
+      ]
+    },
+    {
+      "key": "quality",
+      "title_zh": "证据可信度、反证与方法审计",
+      "title_en": "Evidence Quality, Counterevidence & Method Audit",
+      "lead_zh": "检查证据为什么可信、哪里冲突，以及哪些结论必须降级。",
+      "lead_en": "Examine why evidence is credible, where it conflicts, and which conclusions require downgrading.",
+      "modules": [
+        "quality",
+        "conflicts",
+        "trace"
+      ]
+    },
+    {
+      "key": "action",
+      "title_zh": "适用范围与教学行动",
+      "title_en": "Applicability & Teaching Action",
+      "lead_zh": "把可外推范围、护栏和教学动作连接到具体证据。",
+      "lead_en": "Connect applicability, guardrails and teaching actions to specific evidence.",
+      "modules": [
+        "applicability",
+        "intervention"
+      ]
+    },
+    {
+      "key": "evaluation",
+      "title_zh": "试点设计、评估与停止条件",
+      "title_en": "Pilot, Evaluation & Stop Conditions",
+      "lead_zh": "用独立学习结果验证试点，并预先写清停止条件。",
+      "lead_en": "Validate the pilot with independent learning outcomes and pre-specified stop conditions.",
+      "modules": [
+        "evaluation"
+      ]
+    },
+    {
+      "key": "sources",
+      "title_zh": "来源、溯源与附录",
+      "title_en": "Sources, Traceability & Appendix",
+      "lead_zh": "保留原始来源、URL、证据 ID 和获取信息，确保可回查。",
+      "lead_en": "Preserve original sources, URLs, evidence IDs and retrieval metadata for auditability.",
+      "modules": [
+        "sources"
+      ]
+    }
+  ],
+  "brief_blocks": {
+    "zh": {
+      "decision": [
+        "先看结论",
+        "该不该做、置信度多高、最关键的证据边界在哪。"
+      ],
+      "lieflat": [
+        "Lieflat 实证手作画廊",
+        "AI 按数据形状从 Lieflat 目录选型编排；每张图的数字都可溯源到 result.json。"
+      ],
+      "outcomes": [
+        "任务表现 ≠ 学习效果",
+        "只展示真正有解释力的结果分离；正向、负向与零效应按 effect_direction 编码。"
+      ],
+      "tribunal": [
+        "证据裁决",
+        "支持、不确定、被反驳与缺失证据分开放置，不把长段落平铺在同一层。"
+      ],
+      "action": [
+        "从证据到行动",
+        "适用性、护栏、停止条件与评价连成一条可执行路径。"
+      ],
+      "sources": [
+        "关键来源",
+        "摘要页只列最关键的来源；完整溯源在完整报告中展开。"
+      ]
+    },
+    "en": {
+      "decision": [
+        "Decision first",
+        "What to do, how confident we are, and the most important evidence boundary."
+      ],
+      "lieflat": [
+        "Lieflat Editorial Gallery",
+        "Charts selected and composed by AI from the Lieflat catalog; every number traces back to result.json."
+      ],
+      "outcomes": [
+        "Task performance ≠ learning",
+        "Only informative outcome separation; positive, negative and null effects use effect_direction."
+      ],
+      "tribunal": [
+        "Evidence tribunal",
+        "Supported, uncertain, contradicted and missing evidence stay separated instead of flattened into long prose."
+      ],
+      "action": [
+        "Evidence to action",
+        "Applicability, guardrails, stop conditions and evaluation form one executable path."
+      ],
+      "sources": [
+        "Key sources",
+        "Only the key sources in the brief; full traceability expands in the full report."
+      ]
+    }
+  }
+}

package/domains/education/copy/terminology.json

@@ -0,0 +1,70 @@
+{
+  "domain": "education",
+  "purpose": "Domain terminology: methodology item labels + display names for internal fields.",
+  "methodology_labels": {
+    "zh": {
+      "control_group": "对照组",
+      "randomization": "随机分配",
+      "pre_test": "前测",
+      "post_test": "后测",
+      "retention_test": "保持测试",
+      "transfer_test": "迁移测试",
+      "sample_bias": "样本偏差",
+      "self_selection": "自我选择偏差",
+      "measurement_validity": "测量效度",
+      "confounders": "混杂因素",
+      "instructor_effect": "教师效应",
+      "novelty_effect": "新奇效应",
+      "tool_version_effect": "工具版本效应",
+      "ai_usage_policy": "AI 使用规则",
+      "dropout": "样本流失"
+    },
+    "en": {
+      "control_group": "Control group",
+      "randomization": "Randomization",
+      "pre_test": "Pre-test",
+      "post_test": "Post-test",
+      "retention_test": "Retention test",
+      "transfer_test": "Transfer test",
+      "sample_bias": "Sample bias",
+      "self_selection": "Self-selection bias",
+      "measurement_validity": "Measurement validity",
+      "confounders": "Confounders",
+      "instructor_effect": "Instructor effect",
+      "novelty_effect": "Novelty effect",
+      "tool_version_effect": "Tool-version effect",
+      "ai_usage_policy": "AI usage policy",
+      "dropout": "Attrition"
+    }
+  },
+  "field_labels": {
+    "zh": {
+      "target_learners": "目标学习者",
+      "target_population": "目标人群",
+      "ai_usage_policy": "AI 使用规则",
+      "ai_usage_rule": "AI 规则",
+      "task_vs_learning_guard": "任务 vs 学习护栏",
+      "learning_metrics": "学习指标",
+      "process_metrics": "过程指标",
+      "risk_metrics": "风险指标",
+      "task_performance": "任务表现",
+      "learning_gain": "学习收获",
+      "retention": "保持",
+      "transfer": "迁移"
+    },
+    "en": {
+      "target_learners": "Target learners",
+      "target_population": "Target population",
+      "ai_usage_policy": "AI usage policy",
+      "ai_usage_rule": "AI rule",
+      "task_vs_learning_guard": "Task vs learning guard",
+      "learning_metrics": "Learning metrics",
+      "process_metrics": "Process metrics",
+      "risk_metrics": "Risk metrics",
+      "task_performance": "Task performance",
+      "learning_gain": "Learning gain",
+      "retention": "Retention",
+      "transfer": "Transfer"
+    }
+  }
+}

package/domains/education/manifest.json

@@ -4,7 +4,7 @@
   "description": "education 域只是“指向现有契约”的注册：不新增任何逻辑路径，不引入新 schema / 新校验器 / 新方法学。引擎层 load_domain('education') 只做契约存在性校验；领域选择由主 agent 接 CLI 完成。",
   "contracts_source": {
     "frame_schema": "schemas/education-frame.schema.json（EducationResearchFrame，顶层 additionalProperties:false，required: question/decision_target）",
-    "outcome_taxonomy": "20 token 取自 schemas/evidence.schema.json 的 outcome_type enum；四类映射参考 engine/pilot.py _OUTCOME_CATEGORY",
+    "outcome_taxonomy": "20 token 与四类映射（learning/task_performance/process/risk）都在 domains/education/outcome_taxonomy.json；唯一读取者为 engine/taxonomy.py（未知 token fail-closed）",
     "methodology_checklist": "15 项取自 skill/agents/method-reviewer.md；每项 status 枚举 met/partial/missing/not_applicable",
     "golds_dir": "benchmarks/annotations（30 条人工金标 gold-Q*.json）",
     "references_dir": "references（11 篇教育证据方法学文档）"

package/domains/education/outcome_taxonomy.json

@@ -1,8 +1,8 @@
 {
   "domain": "education",
   "source": {
-    "tokens": "schemas/evidence.schema.json -> outcome_type enum（20 token）",
-    "categories": "engine/pilot.py -> _OUTCOME_CATEGORY（learning/task_performance/process/risk 四类）"
+    "authority": "engine/taxonomy.py",
+    "note": "Sole reader of this contract. Token set and token-to-category mapping live in this file; callers must use engine/taxonomy.py (fail-closed) and must not keep private copies of either table."
   },
   "categories": {
     "learning": {

package/domains/manifest.json

@@ -5,7 +5,7 @@
     {
       "id": "education",
       "name": "education",
-      "description": "教育研究领域（注册域，零逻辑路径）：框架复用 schemas/education-frame.schema.json（EducationResearchFrame），结果分类复用 evidence.schema.json 的 20 个 outcome_type token 并按 engine/pilot.py _OUTCOME_CATEGORY 归为四类，方法学清单复用 skill/agents/method-reviewer.md 的 15 项审查清单。",
+      "description": "教育研究领域（注册域，零逻辑路径）：框架复用 schemas/education-frame.schema.json（EducationResearchFrame），结果分类（20 token 与四类映射）由 domains/education/outcome_taxonomy.json 声明、engine/taxonomy.py 单一读取（未知 token fail-closed），方法学清单复用 skill/agents/method-reviewer.md 的 15 项审查清单。",
       "frame_schema": "schemas/education-frame.schema.json",
       "outcome_taxonomy": "domains/education/outcome_taxonomy.json",
       "methodology_checklist": "domains/education/manifest.json#/methodology_checklist",

package/domains/policy/copy/few_shots.json

@@ -0,0 +1,22 @@
+{
+  "domain": "policy",
+  "purpose": "One written example per decision state, policy voice. No teaching vocabulary; intervention is 干预方案.",
+  "states": {
+    "adopt": {
+      "zh": "独立效果、成本与公平证据一致为正，实施风险可控 → 在目标辖区采用该干预方案，并保留持续监测与退出条款。",
+      "en": "Independent effect, cost and equity evidence is consistently positive with controllable implementation risk → adopt the intervention plan in the target jurisdiction, with ongoing monitoring and exit clauses."
+    },
+    "pilot": {
+      "zh": "试点期产出改善明确，但独立效果与成本证据仍薄 → 有界、分阶段、带对照与成本核算的试点，而不是全面铺开。",
+      "en": "Pilot-period output gains are clear, but independent effect and cost evidence is still thin → a bounded, phased pilot with comparison and cost accounting rather than full rollout."
+    },
+    "reject": {
+      "zh": "已观察到实施风险或分配恶化，且无补偿性目标效果证据 → 不建议采用，并写明受影响人群与补救边界。",
+      "en": "Implementation risk or distributional harm is already observed with no compensatory target-effect evidence → do not adopt, and state affected populations and the remediation boundary."
+    },
+    "insufficient_evidence": {
+      "zh": "现有研究未测量独立效果、成本或分配，或辖区与目标人群错位 → 证据不足，先补关键测量设计，再谈决策。",
+      "en": "Existing studies do not measure independent effects, cost or distribution, or the jurisdiction mismatches the target population → insufficient evidence; fill the key measurement design before deciding."
+    }
+  }
+}

package/domains/policy/copy/framing_enums.json

@@ -0,0 +1,94 @@
+{
+  "domain": "policy",
+  "purpose": "policy frame enum display labels (framing_lexicon part 2).",
+  "frame_enums": {
+    "zh": {
+      "policy_decision": "政策决策",
+      "evidence_review": "证据评审",
+      "pilot_design": "试点设计",
+      "evaluation_design": "评价设计",
+      "primary": "主要结果",
+      "secondary": "次要结果",
+      "risk": "风险结果",
+      "adopt": "采纳",
+      "modify": "修改",
+      "terminate": "终止",
+      "maintain": "维持",
+      "evaluate_impact": "评估影响",
+      "rct": "随机对照试验",
+      "quasi_experimental": "准实验",
+      "observational": "观察性研究",
+      "meta_analysis": "元分析",
+      "systematic_review": "系统综述",
+      "qualitative": "质性研究",
+      "mixed_methods": "混合方法",
+      "survey": "调查研究",
+      "case_study": "案例研究",
+      "literature_review": "文献综述",
+      "online": "线上",
+      "offline": "线下",
+      "hybrid": "混合",
+      "policy_effectiveness": "政策有效性",
+      "cost_effectiveness": "成本效益",
+      "equity": "公平性",
+      "feasibility": "可行性",
+      "implementation_risk": "实施风险",
+      "institutional_reform": "机构改革",
+      "human_supervised_customer_support_assistant": "人工监督的客服助手",
+      "approved_knowledge_base_and_agent_review": "已批准知识库 + 坐席复核",
+      "enterprise_customer_support_staff": "企业客服人员",
+      "enterprise_customer_support": "企业客户支持",
+      "autonomous_agents_and_high_stakes_specialist_advice": "自主代理与高风险专业建议",
+      "under_design_pending_evidence_review": "设计中（待证据评审）",
+      "knowledge_work": "知识型工作",
+      "software_development": "软件开发",
+      "daily": "每日",
+      "weekly": "每周",
+      "monthly": "每月"
+    },
+    "en": {
+      "policy_decision": "Policy decision",
+      "evidence_review": "Evidence review",
+      "pilot_design": "Pilot design",
+      "evaluation_design": "Evaluation design",
+      "primary": "Primary outcomes",
+      "secondary": "Secondary outcomes",
+      "risk": "Risk outcomes",
+      "adopt": "Adopt",
+      "modify": "Modify",
+      "terminate": "Terminate",
+      "maintain": "Maintain",
+      "evaluate_impact": "Evaluate impact",
+      "rct": "Randomised controlled trial",
+      "quasi_experimental": "Quasi-experimental",
+      "observational": "Observational",
+      "meta_analysis": "Meta-analysis",
+      "systematic_review": "Systematic review",
+      "qualitative": "Qualitative",
+      "mixed_methods": "Mixed methods",
+      "survey": "Survey",
+      "case_study": "Case study",
+      "literature_review": "Literature review",
+      "online": "Online",
+      "offline": "Offline",
+      "hybrid": "Hybrid",
+      "policy_effectiveness": "Policy effectiveness",
+      "cost_effectiveness": "Cost-effectiveness",
+      "equity": "Equity",
+      "feasibility": "Feasibility",
+      "implementation_risk": "Implementation risk",
+      "institutional_reform": "Institutional reform",
+      "human_supervised_customer_support_assistant": "Human-supervised customer-support assistant",
+      "approved_knowledge_base_and_agent_review": "Approved knowledge base + agent review",
+      "enterprise_customer_support_staff": "Customer-support staff",
+      "enterprise_customer_support": "Enterprise customer support",
+      "autonomous_agents_and_high_stakes_specialist_advice": "Autonomous agents and high-stakes specialist advice",
+      "under_design_pending_evidence_review": "Under design (pending evidence review)",
+      "knowledge_work": "Knowledge work",
+      "software_development": "Software development",
+      "daily": "Daily",
+      "weekly": "Weekly",
+      "monthly": "Monthly"
+    }
+  }
+}

package/domains/policy/copy/framing_lexicon.json

@@ -0,0 +1,174 @@
+{
+  "domain": "policy",
+  "purpose": "policy frame field/subfield labels (framing_lexicon part 1).",
+  "frame_fields": {
+    "zh": {
+      "question": "研究问题",
+      "decision_object": "决策对象",
+      "intervention": "干预方案",
+      "population": "目标人群",
+      "stakeholders": "利益相关者",
+      "outcomes": "结果构念",
+      "context": "政策情境",
+      "scope": "研究范围",
+      "comparison": "比较条件",
+      "success": "决策成功条件",
+      "success_condition": "决策成功条件",
+      "learner": "目标人群",
+      "course": "适用情境"
+    },
+    "en": {
+      "question": "Research question",
+      "decision_object": "Decision object",
+      "intervention": "Intervention plan",
+      "population": "Target population",
+      "stakeholders": "Stakeholders",
+      "outcomes": "Outcome constructs",
+      "context": "Policy context",
+      "scope": "Research scope",
+      "comparison": "Comparison",
+      "success": "Decision success condition",
+      "success_condition": "Decision success condition",
+      "learner": "Target population",
+      "course": "Operating context"
+    }
+  },
+  "frame_subfields": {
+    "zh": {
+      "learner": {
+        "education_level": "覆盖层级",
+        "major": "岗位序列",
+        "prior_knowledge": "先验经验",
+        "special_characteristics": "人群特征"
+      },
+      "course": {
+        "subject": "业务域",
+        "course_type": "组织形态",
+        "duration": "实施周期"
+      },
+      "intervention": {
+        "teaching_method": "干预方式",
+        "ai_tool": "技术工具",
+        "allowed_usage": "允许使用",
+        "frequency": "使用频率",
+        "duration": "干预周期",
+        "policy_name": "政策名称",
+        "policy_type": "政策类型",
+        "jurisdiction": "适用辖区",
+        "mechanism": "作用机制"
+      },
+      "population": {
+        "definition": "人群定义",
+        "size": "覆盖规模",
+        "vulnerability": "脆弱性"
+      },
+      "scope": {
+        "time_range": "时间范围",
+        "geography": "地域",
+        "study_types": "研究设计"
+      }
+    },
+    "en": {
+      "learner": {
+        "education_level": "Coverage level",
+        "major": "Role track",
+        "prior_knowledge": "Prior experience",
+        "special_characteristics": "Population characteristics"
+      },
+      "course": {
+        "subject": "Domain",
+        "course_type": "Organisation type",
+        "duration": "Rollout window"
+      },
+      "intervention": {
+        "teaching_method": "Method",
+        "ai_tool": "Technology",
+        "allowed_usage": "Allowed usage",
+        "frequency": "Frequency",
+        "duration": "Duration",
+        "policy_name": "Policy",
+        "policy_type": "Policy type",
+        "jurisdiction": "Jurisdiction",
+        "mechanism": "Mechanism"
+      },
+      "population": {
+        "definition": "Definition",
+        "size": "Scale",
+        "vulnerability": "Vulnerability"
+      },
+      "scope": {
+        "time_range": "Time range",
+        "geography": "Geography",
+        "study_types": "Study designs"
+      }
+    }
+  },
+  "evidence_detail_labels": {
+    "zh": {
+      "study_id": "研究 ID",
+      "sample_id": "样本 ID",
+      "title": "研究标题",
+      "year": "年份",
+      "study_type": "研究设计",
+      "education_level": "覆盖层级",
+      "population": "研究人群",
+      "sample_size": "样本量",
+      "intervention": "干预方案",
+      "comparison": "对照 / 比较条件",
+      "outcome_measure": "结果测量",
+      "claim": "完整主张",
+      "effect": "效应 / 结果",
+      "effect_direction": "效应方向",
+      "relation_to_claim": "与主张关系",
+      "duration": "干预时长",
+      "method": "方法",
+      "strengths": "优势",
+      "limitations": "局限",
+      "confounders": "混杂因素",
+      "quality_dimensions": "质量维度",
+      "quality_score": "质量分",
+      "evidence_level": "证据等级",
+      "directness": "直接性",
+      "applicability": "适用性",
+      "confidence": "置信度",
+      "status": "证据状态",
+      "source_location": "来源位置",
+      "source_title": "来源标题",
+      "source_url": "可验证链接",
+      "claim_id": "Claim ID"
+    },
+    "en": {
+      "study_id": "Study ID",
+      "sample_id": "Sample ID",
+      "title": "Study title",
+      "year": "Year",
+      "study_type": "Study design",
+      "education_level": "Coverage level",
+      "population": "Population",
+      "sample_size": "Sample size",
+      "intervention": "Intervention plan",
+      "comparison": "Comparison",
+      "outcome_measure": "Outcome measure",
+      "claim": "Full claim",
+      "effect": "Effect / result",
+      "effect_direction": "Effect direction",
+      "relation_to_claim": "Relation to claim",
+      "duration": "Duration",
+      "method": "Method",
+      "strengths": "Strengths",
+      "limitations": "Limitations",
+      "confounders": "Confounders",
+      "quality_dimensions": "Quality dimensions",
+      "quality_score": "Quality score",
+      "evidence_level": "Evidence level",
+      "directness": "Directness",
+      "applicability": "Applicability",
+      "confidence": "Confidence",
+      "status": "Evidence status",
+      "source_location": "Source location",
+      "source_title": "Source title",
+      "source_url": "Verifiable URL",
+      "claim_id": "Claim ID"
+    }
+  }
+}