diff --git a/backend/tools/review_structured_acceptance_gate.py b/backend/tools/review_structured_acceptance_gate.py index 7c6d33fa..897d9998 100644 --- a/backend/tools/review_structured_acceptance_gate.py +++ b/backend/tools/review_structured_acceptance_gate.py @@ -22,6 +22,7 @@ class StructuredReviewCase: expected_contract_type: str expected_clause_types: List[str] expected_missing_sections: List[str] + batch_id: str = "B-000" source_origin: str = "deidentified_realistic" redaction_level: str = "sanitized" expected_amount_contains: Optional[str] = None @@ -345,6 +346,9 @@ CASES = [ ] SAMPLE_MANIFEST_PATH = Path(__file__).resolve().parents[2] / "project-docs" / "structured-review-samples" / "manifest.json" +BASELINE_ROOT = SAMPLE_MANIFEST_PATH.parent / "baselines" +BASELINE_SNAPSHOTS_DIR = BASELINE_ROOT / "snapshots" +BASELINE_HISTORY_JSONL = BASELINE_ROOT / "review-structured-baseline-history.jsonl" def load_cases_from_manifest(manifest_path: Path) -> List[StructuredReviewCase]: @@ -659,6 +663,36 @@ def build_field_baseline_by_contract(summary: Dict) -> List[Dict[str, object]]: return rows +def build_field_baseline_by_batch(summary: Dict) -> List[Dict[str, object]]: + rows: List[Dict[str, object]] = [] + grouped: Dict[str, Dict[str, int]] = {} + for item in summary["cases"]: + result = item.get("result") or {} + contract_structure = result.get("contract_structure") or {} + case = next(case for case in CASES if case.case_id == item["case_id"]) + bucket = grouped.setdefault(case.batch_id, {"contract_type_pass": 0, "contract_type_total": 0, "payment_pass": 0, "payment_total": 0}) + bucket["contract_type_total"] += 1 + if contract_structure.get("detected_contract_type") == case.expected_contract_type: + bucket["contract_type_pass"] += 1 + if case.expected_payment_milestones: + bucket["payment_total"] += 1 + if len(contract_structure.get("payment_milestones") or []) >= case.expected_payment_milestones: + bucket["payment_pass"] += 1 + for batch_id, bucket in grouped.items(): + rows.append( + { + "batch_id": batch_id, + "contract_type_passed": bucket["contract_type_pass"], + "contract_type_total": bucket["contract_type_total"], + "contract_type_accuracy": round((bucket["contract_type_pass"] / bucket["contract_type_total"]) * 100, 2) if bucket["contract_type_total"] else 0.0, + "payment_milestone_passed": bucket["payment_pass"], + "payment_milestone_total": bucket["payment_total"], + "payment_milestone_accuracy": round((bucket["payment_pass"] / bucket["payment_total"]) * 100, 2) if bucket["payment_total"] else 0.0, + } + ) + return rows + + def build_relationship_baseline_by_type(summary: Dict) -> List[Dict[str, object]]: rows: List[Dict[str, object]] = [] grouped: Dict[str, Dict[str, int]] = {} @@ -697,6 +731,7 @@ def build_sample_inventory() -> List[Dict[str, object]]: return [ { "case_id": case.case_id, + "batch_id": case.batch_id, "label": case.label, "contract_family": case.expected_contract_type, "source_origin": case.source_origin, @@ -768,6 +803,141 @@ def write_csv(path: Path, rows: List[Dict[str, object]]) -> None: path.write_text("\n".join(lines), encoding="utf-8") +def index_rows(rows: List[Dict[str, object]], key_field: str) -> Dict[str, Dict[str, object]]: + return {str(row[key_field]): row for row in rows if key_field in row} + + +def build_trend_rows( + current_rows: List[Dict[str, object]], + previous_rows: List[Dict[str, object]], + key_field: str, + metric_fields: List[str], +) -> List[Dict[str, object]]: + previous_index = index_rows(previous_rows, key_field) + trend_rows: List[Dict[str, object]] = [] + for row in current_rows: + key = str(row[key_field]) + previous_row = previous_index.get(key, {}) + trend_row: Dict[str, object] = {key_field: key} + for metric_field in metric_fields: + current_value = row.get(metric_field) + previous_value = previous_row.get(metric_field) + trend_row[f"current_{metric_field}"] = current_value + trend_row[f"previous_{metric_field}"] = previous_value if previous_value is not None else "N/A" + if isinstance(current_value, (int, float)) and isinstance(previous_value, (int, float)): + trend_row[f"delta_{metric_field}"] = round(float(current_value) - float(previous_value), 2) + else: + trend_row[f"delta_{metric_field}"] = "N/A" + trend_rows.append(trend_row) + return trend_rows + + +def load_previous_snapshot() -> Optional[Dict[str, object]]: + latest_path = BASELINE_ROOT / "latest-summary.json" + if not latest_path.exists(): + return None + return json.loads(latest_path.read_text(encoding="utf-8")) + + +def persist_baseline_history(summary: Dict[str, object]) -> None: + BASELINE_ROOT.mkdir(parents=True, exist_ok=True) + BASELINE_SNAPSHOTS_DIR.mkdir(parents=True, exist_ok=True) + previous_snapshot = load_previous_snapshot() + generated_at = str(summary["generated_at"]) + safe_stamp = generated_at.replace(":", "-").replace(".", "-") + + snapshot_payload = { + "generated_at": generated_at, + "artifact_dir": summary["artifact_dir"], + "sample_count": len(summary.get("sample_inventory") or []), + "field_baseline": summary.get("field_baseline") or [], + "relationship_baseline": summary.get("relationship_baseline") or [], + "field_baseline_by_contract": summary.get("field_baseline_by_contract") or [], + "field_baseline_by_batch": summary.get("field_baseline_by_batch") or [], + "relationship_baseline_by_type": summary.get("relationship_baseline_by_type") or [], + "sample_inventory": summary.get("sample_inventory") or [], + } + + snapshot_path = BASELINE_SNAPSHOTS_DIR / f"{safe_stamp}-review-structured-baseline-snapshot.json" + snapshot_path.write_text(json.dumps(snapshot_payload, ensure_ascii=False, indent=2), encoding="utf-8") + + history_row = { + "generated_at": generated_at, + "artifact_dir": summary["artifact_dir"], + "snapshot_path": str(snapshot_path), + "sample_count": len(summary.get("sample_inventory") or []), + "field_metric_count": len(summary.get("field_baseline") or []), + "relationship_metric_count": len(summary.get("relationship_baseline") or []), + } + with BASELINE_HISTORY_JSONL.open("a", encoding="utf-8") as fh: + fh.write(json.dumps(history_row, ensure_ascii=False) + "\n") + + latest_path = BASELINE_ROOT / "latest-summary.json" + latest_path.write_text(json.dumps(snapshot_payload, ensure_ascii=False, indent=2), encoding="utf-8") + + write_csv(BASELINE_ROOT / "latest-field-baseline.csv", summary.get("field_baseline") or []) + write_csv(BASELINE_ROOT / "latest-relationship-baseline.csv", summary.get("relationship_baseline") or []) + write_csv(BASELINE_ROOT / "latest-field-baseline-by-contract.csv", summary.get("field_baseline_by_contract") or []) + write_csv(BASELINE_ROOT / "latest-field-baseline-by-batch.csv", summary.get("field_baseline_by_batch") or []) + write_csv(BASELINE_ROOT / "latest-relationship-baseline-by-type.csv", summary.get("relationship_baseline_by_type") or []) + write_csv(BASELINE_ROOT / "latest-sample-inventory.csv", summary.get("sample_inventory") or []) + + field_trend_rows = build_trend_rows( + summary.get("field_baseline") or [], + (previous_snapshot or {}).get("field_baseline") or [], + "metric", + ["accuracy"], + ) + relationship_trend_rows = build_trend_rows( + summary.get("relationship_baseline") or [], + (previous_snapshot or {}).get("relationship_baseline") or [], + "metric", + ["accuracy"], + ) + contract_trend_rows = build_trend_rows( + summary.get("field_baseline_by_contract") or [], + (previous_snapshot or {}).get("field_baseline_by_contract") or [], + "contract_family", + ["contract_type_accuracy", "payment_milestone_accuracy", "relationship_presence_accuracy"], + ) + batch_trend_rows = build_trend_rows( + summary.get("field_baseline_by_batch") or [], + (previous_snapshot or {}).get("field_baseline_by_batch") or [], + "batch_id", + ["contract_type_accuracy", "payment_milestone_accuracy"], + ) + relationship_type_trend_rows = build_trend_rows( + summary.get("relationship_baseline_by_type") or [], + (previous_snapshot or {}).get("relationship_baseline_by_type") or [], + "relationship_type", + ["presence_accuracy", "confidence_accuracy"], + ) + + write_csv(BASELINE_ROOT / "baseline-trend-overall.csv", field_trend_rows) + write_csv(BASELINE_ROOT / "baseline-trend-relationship-overall.csv", relationship_trend_rows) + write_csv(BASELINE_ROOT / "baseline-trend-by-contract.csv", contract_trend_rows) + write_csv(BASELINE_ROOT / "baseline-trend-by-batch.csv", batch_trend_rows) + write_csv(BASELINE_ROOT / "baseline-trend-by-relationship-type.csv", relationship_type_trend_rows) + + trend_summary_md = BASELINE_ROOT / "baseline-trend-summary.md" + lines = [ + "# Structured Baseline Trend Summary", + "", + f"- Latest snapshot: `{generated_at}`", + f"- Sample count: `{len(summary.get('sample_inventory') or [])}`", + f"- Previous snapshot: `{(previous_snapshot or {}).get('generated_at') or 'N/A'}`", + "", + "Generated trend files:", + "", + "1. `baseline-trend-overall.csv`", + "2. `baseline-trend-relationship-overall.csv`", + "3. `baseline-trend-by-contract.csv`", + "4. `baseline-trend-by-batch.csv`", + "5. `baseline-trend-by-relationship-type.csv`", + ] + trend_summary_md.write_text("\n".join(lines), encoding="utf-8") + + def run_gate(api_base: str, artifact_dir: Path, password: str) -> Dict: artifact_dir.mkdir(parents=True, exist_ok=True) client = httpx.Client(timeout=120.0) @@ -813,11 +983,13 @@ def run_gate(api_base: str, artifact_dir: Path, password: str) -> Dict: field_baseline = build_field_baseline(summary) relationship_baseline = build_relationship_baseline(summary) field_baseline_by_contract = build_field_baseline_by_contract(summary) + field_baseline_by_batch = build_field_baseline_by_batch(summary) relationship_baseline_by_type = build_relationship_baseline_by_type(summary) sample_inventory = build_sample_inventory() summary["field_baseline"] = field_baseline summary["relationship_baseline"] = relationship_baseline summary["field_baseline_by_contract"] = field_baseline_by_contract + summary["field_baseline_by_batch"] = field_baseline_by_batch summary["relationship_baseline_by_type"] = relationship_baseline_by_type summary["sample_inventory"] = sample_inventory return summary @@ -842,6 +1014,8 @@ def main() -> int: write_csv(relationship_csv_path, summary.get("relationship_baseline") or []) field_by_contract_csv_path = artifact_dir / "review-structured-field-baseline-by-contract.csv" write_csv(field_by_contract_csv_path, summary.get("field_baseline_by_contract") or []) + field_by_batch_csv_path = artifact_dir / "review-structured-field-baseline-by-batch.csv" + write_csv(field_by_batch_csv_path, summary.get("field_baseline_by_batch") or []) relationship_by_type_csv_path = artifact_dir / "review-structured-relationship-baseline-by-type.csv" write_csv(relationship_by_type_csv_path, summary.get("relationship_baseline_by_type") or []) sample_inventory_csv_path = artifact_dir / "review-structured-sample-inventory.csv" @@ -851,6 +1025,7 @@ def main() -> int: build_baseline_md(summary.get("field_baseline") or [], summary.get("relationship_baseline") or []), encoding="utf-8", ) + persist_baseline_history(summary) print(f"REVIEW_STRUCTURED_ACCEPTANCE_{summary['status']} {artifact_dir}") return 0 if summary["status"] == "PASS" else 1 diff --git a/project-docs/OCR后续能力扩展阶段1第八批执行记录-2026-05-04.md b/project-docs/OCR后续能力扩展阶段1第八批执行记录-2026-05-04.md new file mode 100644 index 00000000..62ed9c12 --- /dev/null +++ b/project-docs/OCR后续能力扩展阶段1第八批执行记录-2026-05-04.md @@ -0,0 +1,158 @@ +# OCR 后续能力扩展阶段 1 第八批执行记录（2026-05-04） + +## 1. 本批目标 + +基于阶段 1 第七批已经完成的样本独立样本夹与滚动基线入口，继续推进第八批： + +1. 让真实样本按批次正式进入样本库。 +2. 让滚动基线开始自动留存历史快照。 +3. 让滚动基线开始自动输出趋势对比。 + +## 2. 本批改造 + +### 样本库 + +1. [structured-review-samples/manifest.json](D:/ai-contract-platform/project-docs/structured-review-samples/manifest.json) + - 每条样本正式增加 `batch_id` + +2. [structured-review-samples/batches.json](D:/ai-contract-platform/project-docs/structured-review-samples/batches.json) + - 新增批次索引文件 + - 当前批次： + - `B-001` 基础结构化样本批 + - `B-002` 复杂关系规则样本批 + - `B-003` 扩展合同类型样本批 + +3. [真实业务合同批量样本库-2026-05-04.csv](D:/ai-contract-platform/project-docs/真实业务合同批量样本库-2026-05-04.csv) + - 样本总表新增 `batch_id` + +### 门禁工具 + +1. `backend/tools/review_structured_acceptance_gate.py` + - 运行时样本真相源正式切换为： + - `manifest.json` + - `samples/*.txt` + - 新增自动历史留存： + - `review-structured-baseline-history.jsonl` + - `snapshots/*.json` + - 新增自动趋势文件： + - `baseline-trend-overall.csv` + - `baseline-trend-relationship-overall.csv` + - `baseline-trend-by-contract.csv` + - `baseline-trend-by-batch.csv` + - `baseline-trend-by-relationship-type.csv` + - `baseline-trend-summary.md` + - 新增稳定输出： + - `latest-summary.json` + - `latest-field-baseline.csv` + - `latest-relationship-baseline.csv` + - `latest-field-baseline-by-contract.csv` + - `latest-field-baseline-by-batch.csv` + - `latest-relationship-baseline-by-type.csv` + - `latest-sample-inventory.csv` + +2. `backend/tests/test_review_structured_acceptance_gate.py` + - 新增 / 补强： + - manifest 读取测试 + - 按合同类型滚动基线测试 + - 按关系类型滚动基线测试 + +### 文档 + +1. [structured-review-samples/README.md](D:/ai-contract-platform/project-docs/structured-review-samples/README.md) + - 增加样本库、批次和 `baselines/` 用法说明 + +2. [structured-review-samples/baselines/README.md](D:/ai-contract-platform/project-docs/structured-review-samples/baselines/README.md) + - 新增基线目录说明 + +3. [结构化字段与关系准确率基线-2026-05-04.md](D:/ai-contract-platform/project-docs/结构化字段与关系准确率基线-2026-05-04.md) + - 新增按批次滚动基线入口说明 + +## 3. 当前批次结构 + +### B-001 基础结构化样本批 + +- `S-001` +- `S-002` +- `S-003` +- `S-004` + +### B-002 复杂关系规则样本批 + +- `S-005` +- `S-006` +- `S-007` +- `S-008` + +### B-003 扩展合同类型样本批 + +- `S-009` +- `S-010` + +## 4. 验证结果 + +### 本地验证 + +1. 后端： + - `backend\\.venv\\Scripts\\python.exe -m unittest tests.test_review_structured_acceptance_gate tests.test_review_service tests.test_review_service_rules tests.test_reviews_api` + - `PASS` + +2. 前端： + - `frontend\\npm.cmd run build` + - `PASS` + +### 线上验证 + +1. 结构化门禁： + - `project-docs/qa-artifacts/review-structured-acceptance-gate/2026-05-04T19-49-41-392/` + - 结果：`PASS` + +2. PDF 门禁： + - `project-docs/qa-artifacts/pdf-acceptance-gate/2026-05-04T19-49-41-585/` + - 结果：`PASS` + +## 5. 当前新增资产 + +### 历史留存 + +- `project-docs/structured-review-samples/baselines/review-structured-baseline-history.jsonl` +- `project-docs/structured-review-samples/baselines/snapshots/*.json` + +### 最新基线 + +- `project-docs/structured-review-samples/baselines/latest-field-baseline.csv` +- `project-docs/structured-review-samples/baselines/latest-relationship-baseline.csv` +- `project-docs/structured-review-samples/baselines/latest-field-baseline-by-contract.csv` +- `project-docs/structured-review-samples/baselines/latest-field-baseline-by-batch.csv` +- `project-docs/structured-review-samples/baselines/latest-relationship-baseline-by-type.csv` +- `project-docs/structured-review-samples/baselines/latest-sample-inventory.csv` + +### 趋势对比 + +- `project-docs/structured-review-samples/baselines/baseline-trend-overall.csv` +- `project-docs/structured-review-samples/baselines/baseline-trend-relationship-overall.csv` +- `project-docs/structured-review-samples/baselines/baseline-trend-by-contract.csv` +- `project-docs/structured-review-samples/baselines/baseline-trend-by-batch.csv` +- `project-docs/structured-review-samples/baselines/baseline-trend-by-relationship-type.csv` +- `project-docs/structured-review-samples/baselines/baseline-trend-summary.md` + +## 6. 当前结论 + +阶段 1 第八批已经闭环： + +1. 样本库已经具备批次语义。 +2. 基线已经开始自动留存历史快照。 +3. 基线已经开始自动产出趋势对比。 +4. 结构化门禁已经能对当前表现和历史表现做出持续留痕。 + +## 7. 下一步建议 + +下一步最自然的是进入： + +**阶段 1 第九批：真实来源批量扩容与批次趋势可视化** + +建议优先处理： + +1. 再引入一批新的真实脱敏合同样本。 +2. 给每批样本加来源批次说明和业务线标签。 +3. 把趋势结果做成更直观的汇总页或图表。 +4. 开始识别字段准确率的波动点和弱关系热点。 diff --git a/project-docs/structured-review-samples/README.md b/project-docs/structured-review-samples/README.md index 2d58d2bc..f793a737 100644 --- a/project-docs/structured-review-samples/README.md +++ b/project-docs/structured-review-samples/README.md @@ -8,13 +8,19 @@ - 样本元数据清单 - 包含： - 样本编号 + - 批次编号 - 合同族 - 样本来源说明 - 预期字段 - 预期关系 - 预期置信度 +- `batches.json` + - 样本批次说明 + - 描述每个批次的目标、关注点和包含的样本 - `samples/` - 每条样本对应一份独立文本文件 +- `baselines/` + - 结构化门禁的最新基线、副本快照、趋势文件与历史记录 ## 当前样本 @@ -45,3 +51,13 @@ 当前主要被以下工具读取： - `backend/tools/review_structured_acceptance_gate.py` + +## 基线与趋势 + +当结构化门禁运行后，会在 `baselines/` 下生成： + +1. `latest-*.csv / latest-summary.json` +2. `snapshots/*.json` +3. `review-structured-baseline-history.jsonl` +4. `baseline-trend-*.csv` +5. `baseline-trend-summary.md` diff --git a/project-docs/structured-review-samples/baselines/README.md b/project-docs/structured-review-samples/baselines/README.md new file mode 100644 index 00000000..27b06af3 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/README.md @@ -0,0 +1,39 @@ +# Structured Review Baselines + +## 目录用途 + +本目录用于留存结构化审查门禁的历史快照、最新基线与趋势文件。 + +## 文件说明 + +### 最新基线 + +1. `latest-summary.json` +2. `latest-field-baseline.csv` +3. `latest-relationship-baseline.csv` +4. `latest-field-baseline-by-contract.csv` +5. `latest-field-baseline-by-batch.csv` +6. `latest-relationship-baseline-by-type.csv` +7. `latest-sample-inventory.csv` + +### 历史记录 + +1. `review-structured-baseline-history.jsonl` +2. `snapshots/*.json` + +### 趋势文件 + +1. `baseline-trend-overall.csv` +2. `baseline-trend-relationship-overall.csv` +3. `baseline-trend-by-contract.csv` +4. `baseline-trend-by-batch.csv` +5. `baseline-trend-by-relationship-type.csv` +6. `baseline-trend-summary.md` + +## 使用方式 + +1. 运行 `scripts/run-review-structured-acceptance-gate.ps1` +2. 门禁完成后会自动刷新本目录内容 +3. 使用 `latest-*` 查看当前基线 +4. 使用 `baseline-trend-*` 查看与前一次快照的对比 +5. 使用 `review-structured-baseline-history.jsonl` 查看历史留存 diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-by-batch.csv b/project-docs/structured-review-samples/baselines/baseline-trend-by-batch.csv new file mode 100644 index 00000000..f6245cf8 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-by-batch.csv @@ -0,0 +1,4 @@ +batch_id,current_contract_type_accuracy,previous_contract_type_accuracy,delta_contract_type_accuracy,current_payment_milestone_accuracy,previous_payment_milestone_accuracy,delta_payment_milestone_accuracy +B-001,100.0,100.0,0.0,100.0,100.0,0.0 +B-002,100.0,100.0,0.0,100.0,100.0,0.0 +B-003,100.0,100.0,0.0,100.0,100.0,0.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-by-contract.csv b/project-docs/structured-review-samples/baselines/baseline-trend-by-contract.csv new file mode 100644 index 00000000..773e75d6 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-by-contract.csv @@ -0,0 +1,8 @@ +contract_family,current_contract_type_accuracy,previous_contract_type_accuracy,delta_contract_type_accuracy,current_payment_milestone_accuracy,previous_payment_milestone_accuracy,delta_payment_milestone_accuracy,current_relationship_presence_accuracy,previous_relationship_presence_accuracy,delta_relationship_presence_accuracy +service,100.0,100.0,0.0,100.0,100.0,0.0,100.0,100.0,0.0 +procurement,100.0,100.0,0.0,100.0,100.0,0.0,0.0,0.0,0.0 +sales,100.0,100.0,0.0,100.0,100.0,0.0,100.0,100.0,0.0 +lease,100.0,100.0,0.0,0.0,0.0,0.0,100.0,100.0,0.0 +development,100.0,100.0,0.0,100.0,100.0,0.0,100.0,100.0,0.0 +license,100.0,100.0,0.0,100.0,100.0,0.0,100.0,100.0,0.0 +entrustment,100.0,100.0,0.0,100.0,100.0,0.0,100.0,100.0,0.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-by-relationship-type.csv b/project-docs/structured-review-samples/baselines/baseline-trend-by-relationship-type.csv new file mode 100644 index 00000000..178ac59e --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-by-relationship-type.csv @@ -0,0 +1,8 @@ +relationship_type,current_presence_accuracy,previous_presence_accuracy,delta_presence_accuracy,current_confidence_accuracy,previous_confidence_accuracy,delta_confidence_accuracy +payment_depends_on_acceptance,100.0,100.0,0.0,100.0,100.0,0.0 +payment_depends_on_invoice,100.0,100.0,0.0,0.0,0.0,0.0 +liability_cap_conflicts_with_breach,100.0,100.0,0.0,100.0,100.0,0.0 +payment_depends_on_go_live,100.0,100.0,0.0,0.0,0.0,0.0 +payment_after_warranty_period,100.0,100.0,0.0,100.0,100.0,0.0 +liability_cap_conflicts_with_confidentiality,100.0,100.0,0.0,0.0,0.0,0.0 +liability_cap_conflicts_with_ip,100.0,100.0,0.0,100.0,100.0,0.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-overall.csv b/project-docs/structured-review-samples/baselines/baseline-trend-overall.csv new file mode 100644 index 00000000..50f65a8b --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-overall.csv @@ -0,0 +1,9 @@ +metric,current_accuracy,previous_accuracy,delta_accuracy +contract_type,100.0,100.0,0.0 +amount,100.0,100.0,0.0 +term_start,100.0,100.0,0.0 +term_end,100.0,100.0,0.0 +payment_milestones,100.0,100.0,0.0 +clause_types,100.0,100.0,0.0 +intellectual_property_owner,100.0,100.0,0.0 +liability_carve_outs,100.0,100.0,0.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-relationship-overall.csv b/project-docs/structured-review-samples/baselines/baseline-trend-relationship-overall.csv new file mode 100644 index 00000000..05e46c7b --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-relationship-overall.csv @@ -0,0 +1,6 @@ +metric,current_accuracy,previous_accuracy,delta_accuracy +relationship_presence,100.0,100.0,0.0 +relationship_absence,100.0,100.0,0.0 +relationship_confidence,100.0,100.0,0.0 +relationship_high_confidence_expected,100.0,100.0,0.0 +relationship_weak_hint_expected,100.0,100.0,0.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/baseline-trend-summary.md b/project-docs/structured-review-samples/baselines/baseline-trend-summary.md new file mode 100644 index 00000000..806e66d6 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/baseline-trend-summary.md @@ -0,0 +1,13 @@ +# Structured Baseline Trend Summary + +- Latest snapshot: `2026-05-04T11:49:42.278492+00:00` +- Sample count: `10` +- Previous snapshot: `2026-05-04T11:40:35.844773+00:00` + +Generated trend files: + +1. `baseline-trend-overall.csv` +2. `baseline-trend-relationship-overall.csv` +3. `baseline-trend-by-contract.csv` +4. `baseline-trend-by-batch.csv` +5. `baseline-trend-by-relationship-type.csv` \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-field-baseline-by-batch.csv b/project-docs/structured-review-samples/baselines/latest-field-baseline-by-batch.csv new file mode 100644 index 00000000..0e8e15f9 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-field-baseline-by-batch.csv @@ -0,0 +1,4 @@ +batch_id,contract_type_passed,contract_type_total,contract_type_accuracy,payment_milestone_passed,payment_milestone_total,payment_milestone_accuracy +B-001,4,4,100.0,3,3,100.0 +B-002,4,4,100.0,4,4,100.0 +B-003,2,2,100.0,2,2,100.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-field-baseline-by-contract.csv b/project-docs/structured-review-samples/baselines/latest-field-baseline-by-contract.csv new file mode 100644 index 00000000..fc508971 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-field-baseline-by-contract.csv @@ -0,0 +1,8 @@ +contract_family,contract_type_passed,contract_type_total,contract_type_accuracy,payment_milestone_passed,payment_milestone_total,payment_milestone_accuracy,relationship_presence_passed,relationship_presence_total,relationship_presence_accuracy +service,4,4,100.0,4,4,100.0,13,13,100.0 +procurement,1,1,100.0,1,1,100.0,0,0,0.0 +sales,1,1,100.0,1,1,100.0,2,2,100.0 +lease,1,1,100.0,0,0,0.0,1,1,100.0 +development,1,1,100.0,1,1,100.0,2,2,100.0 +license,1,1,100.0,1,1,100.0,1,1,100.0 +entrustment,1,1,100.0,1,1,100.0,2,2,100.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-field-baseline.csv b/project-docs/structured-review-samples/baselines/latest-field-baseline.csv new file mode 100644 index 00000000..de3cb2e7 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-field-baseline.csv @@ -0,0 +1,9 @@ +metric,passed,total,accuracy +contract_type,10,10,100.0 +amount,8,8,100.0 +term_start,6,6,100.0 +term_end,6,6,100.0 +payment_milestones,9,9,100.0 +clause_types,32,32,100.0 +intellectual_property_owner,3,3,100.0 +liability_carve_outs,7,7,100.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-relationship-baseline-by-type.csv b/project-docs/structured-review-samples/baselines/latest-relationship-baseline-by-type.csv new file mode 100644 index 00000000..9e3ab521 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-relationship-baseline-by-type.csv @@ -0,0 +1,8 @@ +relationship_type,presence_passed,presence_total,presence_accuracy,confidence_passed,confidence_total,confidence_accuracy +payment_depends_on_acceptance,7,7,100.0,3,3,100.0 +payment_depends_on_invoice,7,7,100.0,0,0,0.0 +liability_cap_conflicts_with_breach,3,3,100.0,1,1,100.0 +payment_depends_on_go_live,1,1,100.0,0,0,0.0 +payment_after_warranty_period,1,1,100.0,1,1,100.0 +liability_cap_conflicts_with_confidentiality,1,1,100.0,0,0,0.0 +liability_cap_conflicts_with_ip,1,1,100.0,1,1,100.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-relationship-baseline.csv b/project-docs/structured-review-samples/baselines/latest-relationship-baseline.csv new file mode 100644 index 00000000..2d23ac41 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-relationship-baseline.csv @@ -0,0 +1,6 @@ +metric,passed,total,accuracy +relationship_presence,21,21,100.0 +relationship_absence,3,3,100.0 +relationship_confidence,6,6,100.0 +relationship_high_confidence_expected,4,4,100.0 +relationship_weak_hint_expected,2,2,100.0 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-sample-inventory.csv b/project-docs/structured-review-samples/baselines/latest-sample-inventory.csv new file mode 100644 index 00000000..ee73137e --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-sample-inventory.csv @@ -0,0 +1,11 @@ +case_id,batch_id,label,contract_family,source_origin,redaction_level,expected_relationships,expected_payment_milestones +S-001,B-001,服务合同结构化摘要,service,deidentified_realistic,sanitized,2,1 +S-002,B-001,采购合同缺失条款提示,procurement,deidentified_realistic,sanitized,0,1 +S-003,B-001,购销合同编号条款分段,sales,deidentified_realistic,sanitized,2,1 +S-004,B-001,租赁合同责任限制识别,lease,deidentified_realistic,sanitized,1,0 +S-005,B-002,开发合同付款节点拆解,development,deidentified_realistic,sanitized,2,2 +S-006,B-002,复杂实施合同关系深化,service,deidentified_realistic,sanitized,7,4 +S-007,B-002,实施合同责任限制含 carve-out,service,deidentified_realistic,sanitized,3,1 +S-008,B-002,弱提示付款依赖样本,service,deidentified_realistic,sanitized,1,1 +S-009,B-003,软件许可协议归属与许可关系,license,deidentified_realistic,sanitized,1,2 +S-010,B-003,委托服务合同多节点验收关系,entrustment,deidentified_realistic,sanitized,2,3 \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/latest-summary.json b/project-docs/structured-review-samples/baselines/latest-summary.json new file mode 100644 index 00000000..920ac2b3 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/latest-summary.json @@ -0,0 +1,369 @@ +{ + "generated_at": "2026-05-04T11:49:42.278492+00:00", + "artifact_dir": "D:\\ai-contract-platform\\project-docs\\qa-artifacts\\review-structured-acceptance-gate\\2026-05-04T19-49-41-392", + "sample_count": 10, + "field_baseline": [ + { + "metric": "contract_type", + "passed": 10, + "total": 10, + "accuracy": 100.0 + }, + { + "metric": "amount", + "passed": 8, + "total": 8, + "accuracy": 100.0 + }, + { + "metric": "term_start", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "term_end", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "payment_milestones", + "passed": 9, + "total": 9, + "accuracy": 100.0 + }, + { + "metric": "clause_types", + "passed": 32, + "total": 32, + "accuracy": 100.0 + }, + { + "metric": "intellectual_property_owner", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "liability_carve_outs", + "passed": 7, + "total": 7, + "accuracy": 100.0 + } + ], + "relationship_baseline": [ + { + "metric": "relationship_presence", + "passed": 21, + "total": 21, + "accuracy": 100.0 + }, + { + "metric": "relationship_absence", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "relationship_confidence", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "relationship_high_confidence_expected", + "passed": 4, + "total": 4, + "accuracy": 100.0 + }, + { + "metric": "relationship_weak_hint_expected", + "passed": 2, + "total": 2, + "accuracy": 100.0 + } + ], + "field_baseline_by_contract": [ + { + "contract_family": "service", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 13, + "relationship_presence_total": 13, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "procurement", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 0, + "relationship_presence_total": 0, + "relationship_presence_accuracy": 0.0 + }, + { + "contract_family": "sales", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "lease", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 0, + "payment_milestone_total": 0, + "payment_milestone_accuracy": 0.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "development", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "license", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "entrustment", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + } + ], + "field_baseline_by_batch": [ + { + "batch_id": "B-001", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 3, + "payment_milestone_total": 3, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-002", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-003", + "contract_type_passed": 2, + "contract_type_total": 2, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 2, + "payment_milestone_total": 2, + "payment_milestone_accuracy": 100.0 + } + ], + "relationship_baseline_by_type": [ + { + "relationship_type": "payment_depends_on_acceptance", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 3, + "confidence_total": 3, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_invoice", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_breach", + "presence_passed": 3, + "presence_total": 3, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_go_live", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "payment_after_warranty_period", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_confidentiality", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_ip", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + } + ], + "sample_inventory": [ + { + "case_id": "S-001", + "batch_id": "B-001", + "label": "服务合同结构化摘要", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-002", + "batch_id": "B-001", + "label": "采购合同缺失条款提示", + "contract_family": "procurement", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 0, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-003", + "batch_id": "B-001", + "label": "购销合同编号条款分段", + "contract_family": "sales", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-004", + "batch_id": "B-001", + "label": "租赁合同责任限制识别", + "contract_family": "lease", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 0 + }, + { + "case_id": "S-005", + "batch_id": "B-002", + "label": "开发合同付款节点拆解", + "contract_family": "development", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-006", + "batch_id": "B-002", + "label": "复杂实施合同关系深化", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 7, + "expected_payment_milestones": 4 + }, + { + "case_id": "S-007", + "batch_id": "B-002", + "label": "实施合同责任限制含 carve-out", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 3, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-008", + "batch_id": "B-002", + "label": "弱提示付款依赖样本", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-009", + "batch_id": "B-003", + "label": "软件许可协议归属与许可关系", + "contract_family": "license", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-010", + "batch_id": "B-003", + "label": "委托服务合同多节点验收关系", + "contract_family": "entrustment", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 3 + } + ] +} \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/review-structured-baseline-history.jsonl b/project-docs/structured-review-samples/baselines/review-structured-baseline-history.jsonl new file mode 100644 index 00000000..e65d2d49 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/review-structured-baseline-history.jsonl @@ -0,0 +1,2 @@ +{"generated_at": "2026-05-04T11:40:35.844773+00:00", "artifact_dir": "D:\\ai-contract-platform\\project-docs\\qa-artifacts\\review-structured-acceptance-gate\\2026-05-04T19-40-34-945", "snapshot_path": "D:\\ai-contract-platform\\project-docs\\structured-review-samples\\baselines\\snapshots\\2026-05-04T11-40-35-844773+00-00-review-structured-baseline-snapshot.json", "sample_count": 10, "field_metric_count": 8, "relationship_metric_count": 5} +{"generated_at": "2026-05-04T11:49:42.278492+00:00", "artifact_dir": "D:\\ai-contract-platform\\project-docs\\qa-artifacts\\review-structured-acceptance-gate\\2026-05-04T19-49-41-392", "snapshot_path": "D:\\ai-contract-platform\\project-docs\\structured-review-samples\\baselines\\snapshots\\2026-05-04T11-49-42-278492+00-00-review-structured-baseline-snapshot.json", "sample_count": 10, "field_metric_count": 8, "relationship_metric_count": 5} diff --git a/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-40-35-844773+00-00-review-structured-baseline-snapshot.json b/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-40-35-844773+00-00-review-structured-baseline-snapshot.json new file mode 100644 index 00000000..4053e922 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-40-35-844773+00-00-review-structured-baseline-snapshot.json @@ -0,0 +1,369 @@ +{ + "generated_at": "2026-05-04T11:40:35.844773+00:00", + "artifact_dir": "D:\\ai-contract-platform\\project-docs\\qa-artifacts\\review-structured-acceptance-gate\\2026-05-04T19-40-34-945", + "sample_count": 10, + "field_baseline": [ + { + "metric": "contract_type", + "passed": 10, + "total": 10, + "accuracy": 100.0 + }, + { + "metric": "amount", + "passed": 8, + "total": 8, + "accuracy": 100.0 + }, + { + "metric": "term_start", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "term_end", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "payment_milestones", + "passed": 9, + "total": 9, + "accuracy": 100.0 + }, + { + "metric": "clause_types", + "passed": 32, + "total": 32, + "accuracy": 100.0 + }, + { + "metric": "intellectual_property_owner", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "liability_carve_outs", + "passed": 7, + "total": 7, + "accuracy": 100.0 + } + ], + "relationship_baseline": [ + { + "metric": "relationship_presence", + "passed": 21, + "total": 21, + "accuracy": 100.0 + }, + { + "metric": "relationship_absence", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "relationship_confidence", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "relationship_high_confidence_expected", + "passed": 4, + "total": 4, + "accuracy": 100.0 + }, + { + "metric": "relationship_weak_hint_expected", + "passed": 2, + "total": 2, + "accuracy": 100.0 + } + ], + "field_baseline_by_contract": [ + { + "contract_family": "service", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 13, + "relationship_presence_total": 13, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "procurement", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 0, + "relationship_presence_total": 0, + "relationship_presence_accuracy": 0.0 + }, + { + "contract_family": "sales", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "lease", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 0, + "payment_milestone_total": 0, + "payment_milestone_accuracy": 0.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "development", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "license", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "entrustment", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + } + ], + "field_baseline_by_batch": [ + { + "batch_id": "B-001", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 3, + "payment_milestone_total": 3, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-002", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-003", + "contract_type_passed": 2, + "contract_type_total": 2, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 2, + "payment_milestone_total": 2, + "payment_milestone_accuracy": 100.0 + } + ], + "relationship_baseline_by_type": [ + { + "relationship_type": "payment_depends_on_acceptance", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 3, + "confidence_total": 3, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_invoice", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_breach", + "presence_passed": 3, + "presence_total": 3, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_go_live", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "payment_after_warranty_period", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_confidentiality", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_ip", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + } + ], + "sample_inventory": [ + { + "case_id": "S-001", + "batch_id": "B-001", + "label": "服务合同结构化摘要", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-002", + "batch_id": "B-001", + "label": "采购合同缺失条款提示", + "contract_family": "procurement", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 0, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-003", + "batch_id": "B-001", + "label": "购销合同编号条款分段", + "contract_family": "sales", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-004", + "batch_id": "B-001", + "label": "租赁合同责任限制识别", + "contract_family": "lease", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 0 + }, + { + "case_id": "S-005", + "batch_id": "B-002", + "label": "开发合同付款节点拆解", + "contract_family": "development", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-006", + "batch_id": "B-002", + "label": "复杂实施合同关系深化", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 7, + "expected_payment_milestones": 4 + }, + { + "case_id": "S-007", + "batch_id": "B-002", + "label": "实施合同责任限制含 carve-out", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 3, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-008", + "batch_id": "B-002", + "label": "弱提示付款依赖样本", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-009", + "batch_id": "B-003", + "label": "软件许可协议归属与许可关系", + "contract_family": "license", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-010", + "batch_id": "B-003", + "label": "委托服务合同多节点验收关系", + "contract_family": "entrustment", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 3 + } + ] +} \ No newline at end of file diff --git a/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-49-42-278492+00-00-review-structured-baseline-snapshot.json b/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-49-42-278492+00-00-review-structured-baseline-snapshot.json new file mode 100644 index 00000000..920ac2b3 --- /dev/null +++ b/project-docs/structured-review-samples/baselines/snapshots/2026-05-04T11-49-42-278492+00-00-review-structured-baseline-snapshot.json @@ -0,0 +1,369 @@ +{ + "generated_at": "2026-05-04T11:49:42.278492+00:00", + "artifact_dir": "D:\\ai-contract-platform\\project-docs\\qa-artifacts\\review-structured-acceptance-gate\\2026-05-04T19-49-41-392", + "sample_count": 10, + "field_baseline": [ + { + "metric": "contract_type", + "passed": 10, + "total": 10, + "accuracy": 100.0 + }, + { + "metric": "amount", + "passed": 8, + "total": 8, + "accuracy": 100.0 + }, + { + "metric": "term_start", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "term_end", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "payment_milestones", + "passed": 9, + "total": 9, + "accuracy": 100.0 + }, + { + "metric": "clause_types", + "passed": 32, + "total": 32, + "accuracy": 100.0 + }, + { + "metric": "intellectual_property_owner", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "liability_carve_outs", + "passed": 7, + "total": 7, + "accuracy": 100.0 + } + ], + "relationship_baseline": [ + { + "metric": "relationship_presence", + "passed": 21, + "total": 21, + "accuracy": 100.0 + }, + { + "metric": "relationship_absence", + "passed": 3, + "total": 3, + "accuracy": 100.0 + }, + { + "metric": "relationship_confidence", + "passed": 6, + "total": 6, + "accuracy": 100.0 + }, + { + "metric": "relationship_high_confidence_expected", + "passed": 4, + "total": 4, + "accuracy": 100.0 + }, + { + "metric": "relationship_weak_hint_expected", + "passed": 2, + "total": 2, + "accuracy": 100.0 + } + ], + "field_baseline_by_contract": [ + { + "contract_family": "service", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 13, + "relationship_presence_total": 13, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "procurement", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 0, + "relationship_presence_total": 0, + "relationship_presence_accuracy": 0.0 + }, + { + "contract_family": "sales", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "lease", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 0, + "payment_milestone_total": 0, + "payment_milestone_accuracy": 0.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "development", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "license", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 1, + "relationship_presence_total": 1, + "relationship_presence_accuracy": 100.0 + }, + { + "contract_family": "entrustment", + "contract_type_passed": 1, + "contract_type_total": 1, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 1, + "payment_milestone_total": 1, + "payment_milestone_accuracy": 100.0, + "relationship_presence_passed": 2, + "relationship_presence_total": 2, + "relationship_presence_accuracy": 100.0 + } + ], + "field_baseline_by_batch": [ + { + "batch_id": "B-001", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 3, + "payment_milestone_total": 3, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-002", + "contract_type_passed": 4, + "contract_type_total": 4, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 4, + "payment_milestone_total": 4, + "payment_milestone_accuracy": 100.0 + }, + { + "batch_id": "B-003", + "contract_type_passed": 2, + "contract_type_total": 2, + "contract_type_accuracy": 100.0, + "payment_milestone_passed": 2, + "payment_milestone_total": 2, + "payment_milestone_accuracy": 100.0 + } + ], + "relationship_baseline_by_type": [ + { + "relationship_type": "payment_depends_on_acceptance", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 3, + "confidence_total": 3, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_invoice", + "presence_passed": 7, + "presence_total": 7, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_breach", + "presence_passed": 3, + "presence_total": 3, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "payment_depends_on_go_live", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "payment_after_warranty_period", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_confidentiality", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 0, + "confidence_total": 0, + "confidence_accuracy": 0.0 + }, + { + "relationship_type": "liability_cap_conflicts_with_ip", + "presence_passed": 1, + "presence_total": 1, + "presence_accuracy": 100.0, + "confidence_passed": 1, + "confidence_total": 1, + "confidence_accuracy": 100.0 + } + ], + "sample_inventory": [ + { + "case_id": "S-001", + "batch_id": "B-001", + "label": "服务合同结构化摘要", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-002", + "batch_id": "B-001", + "label": "采购合同缺失条款提示", + "contract_family": "procurement", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 0, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-003", + "batch_id": "B-001", + "label": "购销合同编号条款分段", + "contract_family": "sales", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-004", + "batch_id": "B-001", + "label": "租赁合同责任限制识别", + "contract_family": "lease", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 0 + }, + { + "case_id": "S-005", + "batch_id": "B-002", + "label": "开发合同付款节点拆解", + "contract_family": "development", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-006", + "batch_id": "B-002", + "label": "复杂实施合同关系深化", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 7, + "expected_payment_milestones": 4 + }, + { + "case_id": "S-007", + "batch_id": "B-002", + "label": "实施合同责任限制含 carve-out", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 3, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-008", + "batch_id": "B-002", + "label": "弱提示付款依赖样本", + "contract_family": "service", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 1 + }, + { + "case_id": "S-009", + "batch_id": "B-003", + "label": "软件许可协议归属与许可关系", + "contract_family": "license", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 1, + "expected_payment_milestones": 2 + }, + { + "case_id": "S-010", + "batch_id": "B-003", + "label": "委托服务合同多节点验收关系", + "contract_family": "entrustment", + "source_origin": "deidentified_realistic", + "redaction_level": "sanitized", + "expected_relationships": 2, + "expected_payment_milestones": 3 + } + ] +} \ No newline at end of file diff --git a/project-docs/structured-review-samples/batches.json b/project-docs/structured-review-samples/batches.json new file mode 100644 index 00000000..7d14ca8e --- /dev/null +++ b/project-docs/structured-review-samples/batches.json @@ -0,0 +1,22 @@ +{ + "batches": [ + { + "batch_id": "B-001", + "label": "基础结构化样本批", + "focus": "基础字段、缺失条款、编号条款切分", + "sample_ids": ["S-001", "S-002", "S-003", "S-004"] + }, + { + "batch_id": "B-002", + "label": "复杂关系规则样本批", + "focus": "付款分期、质保金、责任限制、carve-out、弱提示关系", + "sample_ids": ["S-005", "S-006", "S-007", "S-008"] + }, + { + "batch_id": "B-003", + "label": "扩展合同类型样本批", + "focus": "许可协议、委托服务合同与滚动基线扩展", + "sample_ids": ["S-009", "S-010"] + } + ] +} diff --git a/project-docs/structured-review-samples/manifest.json b/project-docs/structured-review-samples/manifest.json index f714d819..36496a46 100644 --- a/project-docs/structured-review-samples/manifest.json +++ b/project-docs/structured-review-samples/manifest.json @@ -2,6 +2,7 @@ "samples": [ { "case_id": "S-001", + "batch_id": "B-001", "label": "服务合同结构化摘要", "contract_type": "service", "review_as": "party_a", @@ -20,6 +21,7 @@ }, { "case_id": "S-002", + "batch_id": "B-001", "label": "采购合同缺失条款提示", "contract_type": "procurement", "review_as": "party_a", @@ -35,6 +37,7 @@ }, { "case_id": "S-003", + "batch_id": "B-001", "label": "购销合同编号条款分段", "contract_type": null, "review_as": "party_a", @@ -50,6 +53,7 @@ }, { "case_id": "S-004", + "batch_id": "B-001", "label": "租赁合同责任限制识别", "contract_type": null, "review_as": "party_a", @@ -66,6 +70,7 @@ }, { "case_id": "S-005", + "batch_id": "B-002", "label": "开发合同付款节点拆解", "contract_type": null, "review_as": "party_a", @@ -84,6 +89,7 @@ }, { "case_id": "S-006", + "batch_id": "B-002", "label": "复杂实施合同关系深化", "contract_type": null, "review_as": "party_a", @@ -98,7 +104,15 @@ "expected_term_start": "2026-08-01", "expected_term_end": "2027-08-01", "expected_payment_milestones": 4, - "expected_relationship_types": ["payment_depends_on_acceptance", "payment_depends_on_invoice", "payment_depends_on_go_live", "payment_after_warranty_period", "liability_cap_conflicts_with_breach", "liability_cap_conflicts_with_confidentiality", "liability_cap_conflicts_with_ip"], + "expected_relationship_types": [ + "payment_depends_on_acceptance", + "payment_depends_on_invoice", + "payment_depends_on_go_live", + "payment_after_warranty_period", + "liability_cap_conflicts_with_breach", + "liability_cap_conflicts_with_confidentiality", + "liability_cap_conflicts_with_ip" + ], "expected_relationship_confidence": { "payment_depends_on_acceptance": "high", "payment_after_warranty_period": "high", @@ -108,6 +122,7 @@ }, { "case_id": "S-007", + "batch_id": "B-002", "label": "实施合同责任限制含 carve-out", "contract_type": null, "review_as": "party_a", @@ -130,6 +145,7 @@ }, { "case_id": "S-008", + "batch_id": "B-002", "label": "弱提示付款依赖样本", "contract_type": null, "review_as": "party_a", @@ -149,6 +165,7 @@ }, { "case_id": "S-009", + "batch_id": "B-003", "label": "软件许可协议归属与许可关系", "contract_type": null, "review_as": "party_a", @@ -170,6 +187,7 @@ }, { "case_id": "S-010", + "batch_id": "B-003", "label": "委托服务合同多节点验收关系", "contract_type": null, "review_as": "party_a", diff --git a/project-docs/真实业务合同批量样本库-2026-05-04.csv b/project-docs/真实业务合同批量样本库-2026-05-04.csv index 66e729c7..c0a954ce 100644 --- a/project-docs/真实业务合同批量样本库-2026-05-04.csv +++ b/project-docs/真实业务合同批量样本库-2026-05-04.csv @@ -1,11 +1,11 @@ -sample_id,title,contract_family,source_origin,redaction_level,source_mode,focus,expected_highlights,confidence_focus -S-001,服务合同结构化摘要,service,deidentified_realistic,sanitized,text_contract,基础结构化摘要,合同类型/金额/期限/付款/保密/违约,高置信关系 -S-002,采购合同缺失条款提示,procurement,deidentified_realistic,sanitized,text_contract,缺失条款识别,付款/交付/缺失验收/缺失保密,弱提示关系 -S-003,购销合同编号条款分段,sales,deidentified_realistic,sanitized,numbered_contract,编号条款切分,交货/付款/争议解决,高置信关系 -S-004,租赁合同责任限制识别,lease,deidentified_realistic,sanitized,liability_cap_contract,责任限制识别,交付/责任限制/违约,高置信关系 -S-005,开发合同付款节点拆解,development,deidentified_realistic,sanitized,milestone_contract,付款分期拆解,首期款/尾款/验收依赖/发票依赖,高置信关系 -S-006,复杂实施合同关系深化,service,deidentified_realistic,sanitized,complex_realistic_contract,复杂关系组合,付款分期/质保金/保密/知识产权/责任限制/违约,高置信关系 -S-007,实施合同责任限制含 carve-out,service,deidentified_realistic,sanitized,carveout_contract,carve-out 规则校正,责任限制/保密/知识产权/违约,高置信与反例 -S-008,弱提示付款依赖样本,service,deidentified_realistic,sanitized,weak_hint_contract,弱提示关系校正,付款依赖验收但无独立验收条款,弱提示关系 -S-009,软件许可协议归属与许可关系,license,deidentified_realistic,sanitized,license_contract,许可协议归属识别,许可费/知识产权归乙方/授权甲方使用/责任限制,高置信与反例 -S-010,委托服务合同多节点验收关系,entrustment,deidentified_realistic,sanitized,entrustment_contract,多节点验收关系,预付款/进度款/尾款/验收依赖/争议解决,高置信关系 +sample_id,batch_id,title,contract_family,source_origin,redaction_level,source_mode,focus,expected_highlights,confidence_focus +S-001,B-001,服务合同结构化摘要,service,deidentified_realistic,sanitized,text_contract,基础结构化摘要,合同类型/金额/期限/付款/保密/违约,高置信关系 +S-002,B-001,采购合同缺失条款提示,procurement,deidentified_realistic,sanitized,text_contract,缺失条款识别,付款/交付/缺失验收/缺失保密,弱提示关系 +S-003,B-001,购销合同编号条款分段,sales,deidentified_realistic,sanitized,numbered_contract,编号条款切分,交货/付款/争议解决,高置信关系 +S-004,B-001,租赁合同责任限制识别,lease,deidentified_realistic,sanitized,liability_cap_contract,责任限制识别,交付/责任限制/违约,高置信关系 +S-005,B-002,开发合同付款节点拆解,development,deidentified_realistic,sanitized,milestone_contract,付款分期拆解,首期款/尾款/验收依赖/发票依赖,高置信关系 +S-006,B-002,复杂实施合同关系深化,service,deidentified_realistic,sanitized,complex_realistic_contract,复杂关系组合,付款分期/质保金/保密/知识产权/责任限制/违约,高置信关系 +S-007,B-002,实施合同责任限制含 carve-out,service,deidentified_realistic,sanitized,carveout_contract,carve-out 规则校正,责任限制/保密/知识产权/违约,高置信与反例 +S-008,B-002,弱提示付款依赖样本,service,deidentified_realistic,sanitized,weak_hint_contract,弱提示关系校正,付款依赖验收但无独立验收条款,弱提示关系 +S-009,B-003,软件许可协议归属与许可关系,license,deidentified_realistic,sanitized,license_contract,许可协议归属识别,许可费/知识产权归乙方/授权甲方使用/责任限制,高置信与反例 +S-010,B-003,委托服务合同多节点验收关系,entrustment,deidentified_realistic,sanitized,entrustment_contract,多节点验收关系,预付款/进度款/尾款/验收依赖/争议解决,高置信关系 diff --git a/project-docs/结构化字段与关系准确率基线-2026-05-04.md b/project-docs/结构化字段与关系准确率基线-2026-05-04.md index 999daf23..a563e714 100644 --- a/project-docs/结构化字段与关系准确率基线-2026-05-04.md +++ b/project-docs/结构化字段与关系准确率基线-2026-05-04.md @@ -13,6 +13,7 @@ 3. `review-structured-sample-inventory.csv` 4. `review-structured-baseline.md` 5. `review-structured-field-baseline-by-contract.csv` +6. `review-structured-field-baseline-by-batch.csv` 6. `review-structured-relationship-baseline-by-type.csv` ## 2. 字段准确率基线 @@ -63,7 +64,9 @@ 1. 按合同类型： - `review-structured-field-baseline-by-contract.csv` -2. 按关系类型： +2. 按批次： + - `review-structured-field-baseline-by-batch.csv` +3. 按关系类型： - `review-structured-relationship-baseline-by-type.csv` ## 6. 基线意义