docs: create milestone v1.6 roadmap (3 phases, 30 requirements)

.planning/STATE.md

@@ -3,11 +3,11 @@ gsd_state_version: 1.0
 milestone: v1.6
 milestone_name: 语义分段改进
 status: in_progress
-stopped_at: Defining requirements
-last_updated: "2026-03-21T16:00:00.000Z"
-last_activity: 2026-03-21 — Milestone v1.6 started
+stopped_at: Planning Phase 13
+last_updated: "2026-03-21T16:30:00.000Z"
+last_activity: 2026-03-21 — Roadmap created for v1.6
 progress:
-  total_phases: 0
+  total_phases: 3
   completed_phases: 0
   total_plans: 0
   completed_plans: 0
@@ -25,12 +25,23 @@ See: .planning/PROJECT.md (updated 2026-03-21)
 
 ## Current Position
 
-Phase: Not started (defining requirements)
-Plan: —
-Status: Defining requirements
-Last activity: 2026-03-21 — Milestone v1.6 started
+Phase: Phase 13 - 核心 LLM 分段功能
+Plan: TBD
+Status: Planning
+Last activity: 2026-03-21 — Roadmap created for v1.6
 
-Progress: [░░░░░░░░░░] 0%
+Progress: [░░░░░░░░░░] 0% (Phase 13/15)
+
+### Phase 13 Goal
+实现基于 LLM 的智能语义分段，识别视频类型并生成主题标题
+
+### Phase 13 Requirements (18 total)
+- VIDEO-01, VIDEO-02, VIDEO-03 (视频类型识别)
+- BOUND-01, BOUND-02, BOUND-03, BOUND-04 (语义边界检测)
+- COMP-01, COMP-02, COMP-03 (内容完整性保护)
+- TITLE-01, TITLE-02, TITLE-03 (主题标题生成)
+- SENT-01, SENT-02, SENT-03 (中文句边界检测)
+- STRAT-01, STRAT-02, STRAT-03 (分段策略)
 
 ## Accumulated Context
 
@@ -51,6 +62,38 @@ v1.6 scope (待验证):
 - 主题识别与标题 — 主题章节生成标题
 - 内容完整性 — 避免错误截断话题
 
+### Technical Context
+
+**架构集成:**
+- 扩展现有 `SemanticSegmenter`，保持向后兼容
+- 新增 `SemanticSegmentationEngine` 模块（`kb_workflow/srt/semantic_engine.py`）
+- 在 `SRTProcessor` 中添加 LLM 分支（`enable_semantic_segmentation` 参数）
+
+**技术栈:**
+- SimpleAIClient（已集成 zhipu/openai/minimax）
+- Pydantic v2（LLM 响应验证）
+- 现有 SemanticSegmenter（规则分段保留）
+
+**数据流:**
+```
+SRT Input → Parse Subtitles → Video Type Detection →
+Semantic Segmentation → Extract Paragraphs → ASR Correction →
+Deep Analysis
+```
+
+### Known Risks
+
+**高优先级风险:**
+1. **上下文窗口溢出** — 长视频（>1小时）字幕超10万字符，需分块处理
+2. **LLM 输出格式不稳定** — 需多层 fallback 策略
+3. **分段边界破坏完整性** — 需句边界保护机制
+4. **分段不一致性** — LLM 随机性导致多次运行结果不同
+5. **成本失控** — 高频 LLM 调用可能导致费用激增
+
+**缓解策略:**
+- Phase 13: 实现 Pydantic 验证 + 句边界保护
+- Phase 14: 实现分块处理 + 缓存 + 温度控制
+
 ### Blockers/Concerns
 
 None.
@@ -58,7 +101,7 @@ None.
 ## Session Continuity
 
 Last session: 2026-03-21
-Stopped at: Milestone v1.6 started
+Stopped at: Roadmap created for v1.6
 Resume file: None
 
-**Next step:** Define requirements for semantic segmentation
+**Next step:** `/gsd:plan-phase 13` - 开始 Phase 13 的详细规划