16gagent/research/pr-37/docs/pr-37-domain-intelligence.md

# PR-37 Domain Intelligence & Delivery Planning

> **Version:** 1.0 | **Date:** 2026-06-05 | **Status:** ✅ Complete
>
> 增强 SF-02 的行业知识能力和交付规划能力，使之从通用需求工厂升级为行业级需求工厂。

---

## Architecture Overview

```
                        PR-37
         ┌─────────────────────────────────┐
         │                                 │
         │  ┌─ Domain Intelligence ──────┐ │
SF-01 ──┤  │  A. Domain Registry        │ │
   │     │  │  B. Domain Classifier      │ │
SF-02 ──┤  │  C. Industry Epic Gen      │ │
   │     │  │  D. Industry Feature Gen   │ │
   │     │  │  E. Domain Validator       │ │
   │     │  └────────────────────────────┘ │
   │     │                                 │
   └─────┤  ┌─ Delivery Planning ────────┐ │
         │  │  F. Sprint Planner         │ │
         │  │  G. Release Planner        │ │
         │  │  H. Batch Planner          │ │──► SF-03
         │  │  I. Capacity Planner       │ │
         │  │  J. Roadmap Aligner        │ │
         │  └────────────────────────────┘ │
         │                                 │
         │  K. Factory Integration         │
         │     (PR-35/PR-36/SF-01/SF-02)  │
         └─────────────────────────────────┘
```

## Module Map

| Part | Module | File | Lines |
|------|--------|------|-------|
| A | Domain Registry | `domain-registry.mjs` | 1,278 |
| B | Domain Classifier | `domain-classifier.mjs` | 456 |
| C | Industry Epic Generator | `industry-epic-generator.mjs` | 265 |
| D | Industry Feature Generator | `industry-feature-generator.mjs` | 263 |
| E | Domain Validator | `domain-validator.mjs` | 290 |
| F | Sprint Planner | `sprint-planner.mjs` | 472 |
| G | Release Planner | `release-planner.mjs` | 385 |
| H | Batch Planner | `batch-planner.mjs` | 284 |
| I | Capacity Planner | `capacity-planner.mjs` | 276 |
| J | Roadmap Aligner | `roadmap-aligner.mjs` | 384 |
| — | Index (Domain Int.) | `index.mjs` | 179 |
| — | Index (Delivery) | `index.mjs` | 178 |
| **Total** | | **12 files** | **4,710** |

## Domain Registry (Part A)

16 个行业领域，每个行业 10 个知识维度：

| Domain | 中文名 | Epics | Features | Category |
|--------|--------|-------|----------|----------|
| ERP | 企业资源计划 | 10 | 67 | enterprise-software |
| MES | 制造执行系统 | 10 | 67 | industrial |
| CRM | 客户关系管理 | 9 | 63 | enterprise-software |
| WMS | 仓库管理系统 | 7 | 49 | enterprise-software |
| SCM | 供应链管理 | 7 | 42 | enterprise-software |
| HRM | 人力资源管理 | 8 | 56 | management |
| PLM | 产品生命周期管理 | 7 | 42 | engineering |
| CAD | CAD设计平台 | 8 | 56 | engineering |
| EDA | 电子设计自动化 | 8 | 56 | engineering |
| AI Platform | AI平台 | 8 | 56 | intelligence |
| E-Commerce | 电子商务平台 | 10 | 80 | commerce |
| SaaS | SaaS平台 | 8 | 56 | platform |
| Project Management | 项目管理 | 8 | 56 | management |
| Knowledge Base | 知识管理平台 | 8 | 56 | management |
| BI | 商业智能平台 | 8 | 56 | intelligence |
| IoT Platform | 物联网/工业互联网 | 9 | 72 | industrial |

### Knowledge Dimensions per Domain

每个行业包含：
1. `name` / `displayName` — 中英文名
2. `description` — 行业描述
3. `coreProcesses` — 核心业务流程 (3-8)
4. `coreEntities` — 核心业务实体 (10-30)
5. `coreEpics` — 核心Epic模板 (7-10)
6. `coreFeatures` — 核心Feature模板 (40-80 features)
7. `coreKPIs` — 核心KPI (5-10)
8. `coreRoles` — 核心角色 (5-12)
9. `terminology` — 行业术语表 (5-10)
10. `constraints` — 行业约束
11. `regulations` — 行业法规

## Domain Classification (Part B)

### Algorithm

1. 从产品名称和描述中提取关键词
2. 过滤停用词（系统、平台、管理 等）
3. 与 16 个领域的 `keywords` / `aliases` 匹配
4. 对每个领域计算匹配分：关键词 × 3 + 别名 × 5 + 名称直接匹配 × 10
5. 应用交叉匹配映射（如 "跨境电商" → E-Commerce/SCM）
6. 返回 primaryDomain（最高分）、secondaryDomain（次高分）、confidenceScore

### Examples

| Input | Primary | Secondary | Confidence |
|-------|---------|-----------|------------|
| AI客服平台 | AI Platform | E-Commerce | 75% |
| ERP系统 | ERP | — | 48% |
| 工业MES系统 | MES | — | 75% |
| 跨境电商平台 | E-Commerce | SCM | 83% |
| 工业AI质检 | MES | AI Platform | 87% |

## Delivery Planning (Parts F-J)

### Sprint Planning (Part F)

输入 SF-02 的 stories / dependencies / priorities → 输出 Sprint Plan：

```
Sprint 1: Week 1-2, 7 stories, 20 pts, "Sprint 1: 浏览用户注册列表 / ..."
Sprint 2: Week 3-4, 11 stories, 25 pts, "Sprint 2: ..."
Sprint 3: Week 5-6, 8 stories, 20 pts, "Sprint 3: ..."
```

- 基于拓扑排序的依赖约束（依赖的先执行）
- Sprint 容量控制（velocity / maxStoryPointsPerSprint）
- 自动计算 metrics（totalSprints, avgPointsPerSprint, estimatedDurationWeeks）

### Release Planning (Part G)

MVP/V1/V2/Enterprise/Global → 5 个 Release：

```
Release 1.0.0 (MVP): 10 stories, 验证门禁 x5
Release 1.1.0 (MVP Polish): important patches
Release 2.0.0 (V1): 20 stories
Release 3.0.0 (V2): 30 stories
Release 4.0.0 (Global): Global features
```

每个 Release 包含：features, validationGates, releaseGoals, dependencyCheck, releaseDate

### Batch Planning (Part H)

按依赖层级分组：
- **DevelopmentBatch L0** — 无依赖，可并行开发
- **DevelopmentBatch L1** — 依赖 L0，第二批开发
- **TestingBatch** — 开发完成的批次进入测试
- **ReleaseBatch** — 测试通过的进入发布
- **DeliveryBatch** — 客户交付批次

### Capacity Planning (Part I)

- 复杂度因子（high/medium/low）调整工作量
- 团队容量计算：teamSize × velocity × sprintCount × availabilityFactor
- 容量利用率百分比
- 风险缓冲（riskBufferPercent）
- 交付预测（乐观/现实/悲观）

### Roadmap Alignment (Part J)

验证 SF-01 Roadmap 与 PR-37 计划一致性：
- Phase coverage check
- Story coverage check
- Timeline consistency
- Budget alignment
- Dependency integrity
- Alignment Score 0-100

## Factory Integration (Part K)

### createDomainIntelligenceFactory(strategyPackage, options?)

Returns: `{ classification, epics, features, validation, warnings }`

Options: `{ domainOverride, includeValidation }`

### createDeliveryPlanningFactory(upstreamArtifacts, config?)

Accepts SF-02 output as upstreamArtifacts.
Returns: `{ factoryId, status, sprintPlan, releasePlan, batchPlan, capacityPlan, alignmentReport, warnings, metadata }`

Config: `{ sprint: {...}, capacity: {...} }`

## Artifacts (Part L)

```js
PR37_ARTIFACTS = {
  SPRINT_PLAN:      "sprint-plan",
  RELEASE_PLAN:     "release-plan",
  BATCH_PLAN:       "batch-plan",
  CAPACITY_PLAN:    "capacity-plan",
  ALIGNMENT_REPORT: "alignment-report",
}

SF02_ARTIFACTS (新增，from Domain Intelligence):
  DOMAIN_REGISTRY
  INDUSTRY_EPIC_CATALOG
  INDUSTRY_FEATURE_CATALOG
  DOMAIN_VALIDATION_REPORT
```

## Testing (Part M)

57 个测试，12 个 suites，0 个失败：

| Suite | Tests | Coverage |
|-------|-------|----------|
| Part A — Domain Registry | 9 | 16 domains, 10 dims each |
| Part B — Domain Classifier | 9 | All test products + edge cases |
| Part C — Industry Epic Gen | 5 | ERP/MES/AI/fallback |
| Part D — Industry Feature Gen | 4 | ERP features + type detection |
| Part E — Domain Validator | 5 | ERP/MES/generic/unknown |
| Part F — Sprint Planner | 6 | Capacity, deps, edge cases |
| Part G — Release Planner | 4 | 5 releases, gates, null |
| Part H — Batch Planner | 2 | Dep levels, batch types |
| Part I — Capacity Planner | 3 | Effort, forecast, empty |
| Part J — Roadmap Aligner | 1 | Full alignment |
| Part K — Factory Integration | 5 | DI + DP factories |
| Part Z — Regression | 4 | SF-01, SF-02, pipeline, topo |
| **Total** | **57** | **0 failures** |

### Regression Suite

- SF-01: 77 tests → all pass
- SF-02: 89 tests → all pass
- PR-37: 57 tests → all pass
- **Total: 223 tests, 0 failures**

## Quick Start

```js
import { classifyProduct, createDomainIntelligenceFactory } from './src/pr-37-domain-intelligence/index.mjs';
import { createDeliveryPlanningFactory } from './src/pr-37-delivery-planning/index.mjs';

// 1. Classify product
const classification = classifyProduct({ productName: 'ERP系统' });
console.log(classification.primaryDomain); // "ERP"

// 2. Generate industry-specific epics and features
const diFactory = createDomainIntelligenceFactory({ productName: 'ERP系统' });
console.log(diFactory.epics.length);       // 10 (vs 6 generic)
console.log(diFactory.features.length);     // 67 (vs 21 generic)
console.log(diFactory.validation.overallVerdict); // "pass"

// 3. Generate delivery plans from SF-02 output
const dpFactory = createDeliveryPlanningFactory(sf02Output, {
  sprint: { teamVelocity: 20 },
  capacity: { teamSize: 5 },
});
console.log(dpFactory.sprintPlan.sprints.length);     // e.g., 6
console.log(dpFactory.releasePlan.releases.length);   // 5
console.log(dpFactory.alignmentReport.overallVerdict); // "aligned" / "partial"
```

## Files

```
src/pr-37-domain-intelligence/
├── domain-registry.mjs          # Part A: 16 industry domain definitions
├── domain-classifier.mjs        # Part B: product-to-domain classification
├── industry-epic-generator.mjs  # Part C: domain-specific epic generation
├── industry-feature-generator.mjs # Part D: domain-specific feature generation
├── domain-validator.mjs         # Part E: domain coverage validation
└── index.mjs                    # Barrel + createDomainIntelligenceFactory()

src/pr-37-delivery-planning/
├── sprint-planner.mjs           # Part F: Sprint plan from stories + deps
├── release-planner.mjs          # Part G: Release plan from MVP scope
├── batch-planner.mjs            # Part H: Dev/Test/Release/Delivery batches
├── capacity-planner.mjs         # Part I: Team capacity + delivery forecast
├── roadmap-aligner.mjs          # Part J: Sprint/Release/Roadmap alignment
└── index.mjs                    # Barrel + createDeliveryPlanningFactory()

test/
└── pr-37-domain-intelligence-delivery-planning.test.mjs  # 57 tests

docs/pr-37/
├── pr-37-domain-intelligence.md   # This doc
└── pr-37-delivery-planning.md     # Delivery planning details
```