diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..ccc203a
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,62 @@
+"""共享测试 fixtures 和配置"""
+
+from typing import Optional
+from unittest.mock import Mock
+
+import pytest
+
+
+class MockAIClient:
+    """模拟 AI 客户端"""
+
+    def __init__(self, response: str = "Mock AI response"):
+        self.response = response
+
+    def generate(
+        self,
+        prompt: str,
+        max_tokens: int = 16000,
+        disable_reasoning: Optional[bool] = None
+    ) -> str:
+        """生成模拟响应"""
+        return self.response
+
+
+@pytest.fixture
+def mock_ai_client():
+    """提供模拟 AI 客户端"""
+    return MockAIClient()
+
+
+@pytest.fixture
+def sample_segments():
+    """提供示例字幕分段数据"""
+    return [
+        {"index": 1, "start": "00:00:00,000", "end": "00:00:05,000", "text": "大家好，欢迎来到我的频道"},
+        {"index": 2, "start": "00:00:05,000", "end": "00:00:10,000", "text": "今天我们要讲的是机器学习的基础知识"},
+        {"index": 3, "start": "00:00:10,000", "end": "00:00:15,000", "text": "首先，什么是机器学习"},
+        {"index": 4, "start": "00:00:15,000", "end": "00:00:20,000", "text": "机器学习是人工智能的一个分支"},
+        {"index": 5, "start": "00:00:20,000", "end": "00:00:25,000", "text": "它让计算机能够从数据中学习"},
+    ]
+
+
+@pytest.fixture
+def sample_metadata():
+    """提供示例视频元数据"""
+    return {
+        "title": "机器学习入门教程",
+        "category": "教育",
+        "tags": ["编程", "AI", "教程"],
+        "description": "从零开始学习机器学习"
+    }
+
+
+@pytest.fixture
+def sample_metadata_movie():
+    """提供电影类型示例元数据"""
+    return {
+        "title": "星际穿越",
+        "category": "影视",
+        "tags": ["科幻", "电影"],
+        "description": "一部关于太空探索的科幻电影"
+    }
diff --git a/tests/test_semantic_engine.py b/tests/test_semantic_engine.py
new file mode 100644
index 0000000..928dcf3
--- /dev/null
+++ b/tests/test_semantic_engine.py
@@ -0,0 +1,129 @@
+"""测试语义分段引擎"""
+
+import pytest
+
+from kb_workflow.srt.semantic_engine import (
+    VideoSegment,
+    SegmentationResult,
+    SemanticSegmentationEngine
+)
+
+
+class TestVideoTypeDetection:
+    """测试视频类型识别功能"""
+
+    def test_video_type_detection_from_title(self):
+        """测试从标题推断视频类型"""
+        # 知识类 - 从标题推断
+        metadata_edu = {"title": "Python编程教程", "category": "教育"}
+        engine = SemanticSegmentationEngine()
+        video_type = engine._infer_video_type(metadata_edu)
+        assert video_type == "知识类"
+
+        # 知识类 - 包含"课程"
+        metadata_course = {"title": "机器学习课程", "category": ""}
+        video_type = engine._infer_video_type(metadata_course)
+        assert video_type == "知识类"
+
+    def test_video_type_detection_from_category(self):
+        """测试从分类推断视频类型"""
+        # 故事类 - 从分类推断
+        metadata_movie = {"title": "星际穿越", "category": "影视"}
+        engine = SemanticSegmentationEngine()
+        video_type = engine._infer_video_type(metadata_movie)
+        assert video_type == "故事类"
+
+        # 访谈类 - 从标签推断
+        metadata_interview = {"title": "名人访谈", "tags": ["访谈", "对话"]}
+        video_type = engine._infer_video_type(metadata_interview)
+        assert video_type == "访谈类"
+
+    def test_video_type_fallback_to_default(self):
+        """测试元数据缺失时使用默认类型"""
+        # 空元数据
+        engine = SemanticSegmentationEngine()
+        video_type = engine._infer_video_type({})
+        assert video_type == "知识类"  # 默认类型
+
+        # 无效元数据
+        video_type = engine._infer_video_type({"title": "", "category": ""})
+        assert video_type == "知识类"
+
+
+class TestVideoTypeInPrompt:
+    """测试视频类型在 Prompt 中的使用"""
+
+    def test_video_type_in_prompt(self):
+        """测试 Prompt 包含视频类型信息"""
+        engine = SemanticSegmentationEngine()
+        metadata = {"title": "机器学习教程", "category": "教育"}
+
+        prompt = engine._build_prompt("测试内容", metadata)
+
+        assert "视频类型" in prompt or "知识类" in prompt
+        assert "机器学习教程" in prompt
+
+    def test_prompt_format(self):
+        """测试 Prompt 格式正确"""
+        engine = SemanticSegmentationEngine()
+        content = "这是测试内容"
+        metadata = {"title": "测试视频"}
+
+        prompt = engine._build_prompt(content, metadata)
+
+        assert "请分析以下视频字幕" in prompt or "字幕内容" in prompt
+        assert content in prompt
+
+
+class TestPydanticModels:
+    """测试 Pydantic 模型验证"""
+
+    def test_video_segment_model(self):
+        """测试 VideoSegment 模型"""
+        # 创建完整分段
+        segment = VideoSegment(
+            title="机器学习概述",
+            content="机器学习是人工智能的一个分支",
+            start_index=1,
+            end_index=5
+        )
+        assert segment.title == "机器学习概述"
+        assert segment.content == "机器学习是人工智能的一个分支"
+        assert segment.start_index == 1
+        assert segment.end_index == 5
+
+        # 创建无标题分段（title 可选）
+        segment_no_title = VideoSegment(
+            content="测试内容",
+            start_index=1,
+            end_index=3
+        )
+        assert segment_no_title.title is None
+
+    def test_segmentation_result_model(self):
+        """测试 SegmentationResult 模型"""
+        segments = [
+            VideoSegment(title="第一章", content="内容1", start_index=1, end_index=5),
+            VideoSegment(title="第二章", content="内容2", start_index=6, end_index=10)
+        ]
+
+        result = SegmentationResult(
+            video_type="知识类",
+            segments=segments,
+            actual_segment_count=2
+        )
+
+        assert result.video_type == "知识类"
+        assert len(result.segments) == 2
+        assert result.actual_segment_count == 2
+
+    def test_model_validation(self):
+        """测试模型字段验证"""
+        # 缺少必填字段应该抛出异常
+        with pytest.raises(Exception):  # Pydantic ValidationError
+            VideoSegment(
+                title="测试",
+                # 缺少 content
+                start_index=1,
+                end_index=3
+            )