Files
llm_wiki/tools/experiments/wiki-generation-compare/prompts/twostep-analysis.md
T

100 lines
2.3 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
你是一位知识库分析专家。请仔细分析以下源文件,提取结构化信息。
## 任务说明
从源文件中提取以下结构化信息,以 JSON 格式输出:
```json
{
"source_info": {
"filename": "文件名",
"total_lines": ,
"total_words": ,
"main_sections": ["主要小节1", "主要小节2", ...]
},
"concepts": [
{
"term": "概念名称",
"definition": "定义内容",
"line_number": ,
"occurrences": ,
"summary": "简要概述"
}
],
"methods": [
{
"term": "方法名称",
"definition": "定义内容",
"line_number": ,
"occurrences": ,
"steps": ["步骤1", "步骤2", ...]
}
],
"entities": [
{
"term": "实体名称",
"definition": "定义内容",
"line_number": ,
"occurrences": ,
"location": "位置描述"
}
],
"relationships": [
{
"source": "术语A",
"target": "术语B",
"type": "part_of|extends|example_of|contradicts|supports",
"description": "关系说明"
}
],
"key_data": [
{
"type": "数字|百分比|结论",
"content": "数据内容",
"line_number":
}
],
"source_summary": "源文件的完整摘要,约 200-300 字",
"recommended_pages": [
{
"title": "页面标题",
"type": "concept|entity|method",
"reason": "创建理由"
}
]
}
```
## 分析要求
1. **术语识别**
- 概念类:理论术语(如:精气神、天人感应)
- 方法类:修炼方法(如:正身、开窍)
- 实体类:关窍位置(如:泥丸、意窍)
2. **定义提取**
- 优先提取书中明确定义的句子
- 记录行号(从 1 开始计数)
3. **关系识别**
- 识别术语间的依赖、包含、示例、矛盾等关系
- 特别关注"部分"关系(如:精气神包含精、气、神)
4. **数据点提取**
- 数字、百分比、具体结论
- 必须记录行号
5. **推荐页面**
- 基于分析结果,推荐应创建的 Wiki 页面
- 说明推荐理由
## 源文件内容
{SOURCE_CONTENT}
## 输出要求
- 仅输出 JSON,不要包含其他文字
- 确保是有效的 JSON 格式
- 数组项数量根据实际内容确定,不必强制填满