🏛️ 两会提案 AI 评分

2026 年全国两会人大代表提案，由 DeepSeek R1 从最广大人民群众根本利益的立场进行 1-100 分支持度评分。

✨ Features

• 📋 自动爬取 中国人大网 最新两会提案
• 🤖 DeepSeek R1 四维评分：民生改善 / 可行性 / 紧迫性 / 覆盖面
• 📊 交互式数据可视化：环形图、维度卡片、筛选表格
• ⏱️ GitHub Actions 自动化：每日定时运行 pipeline
• 🌐 Vercel 静态部署：零服务器维护

🔀 Architecture

GitHub Actions (daily cron)
  ├── scripts/pipeline.py     ← 爬取 + 评分
  ├── data/scores.json        ← 结果数据
  └── git push → triggers Vercel rebuild

Vercel (static hosting)
  └── resources/public/       ← HTML + CSS + JS
       └── reads /data/scores.json

🚀 Deploy

1. Fork this repo

2. Add GitHub Secret

Settings → Secrets → Actions → New secret
Name:  SILICONFLOW_API_KEY
Value: your-silicon-flow-api-key

3. Connect to Vercel

Import the repo in Vercel, set Output Directory to resources/public.

4. Trigger first run

Go to Actions → Score NPC Proposals → Run workflow

🛠️ Local Development

# Install dependencies
pip install scrapling httpx
python -c "from scrapling.fetchers import DynamicFetcher; DynamicFetcher.setup()"

# Run pipeline
SILICONFLOW_API_KEY=sk-xxx python scripts/pipeline.py

# Serve static site
cd resources/public && python -m http.server 3000

📂 Project Structure

npc-scorer/
├── .github/workflows/score.yml   # GitHub Actions workflow
├── scripts/
│   ├── pipeline.py               # Unified scrape + score pipeline
│   └── scrape_npc.py             # Standalone scraper (legacy)
├── data/
│   └── scores.json               # Generated scoring results
├── resources/public/             # Static frontend
│   ├── index.html
│   ├── style.css
│   └── app.js
├── vercel.json
└── README.md

⚠️ Disclaimer

AI 评分仅供技术演示和参考，不代表任何官方立场。