# DaChuang

**Repository Path**: palatio/da-chuang

## Basic Information

- **Project Name**: DaChuang
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-03-05
- **Last Updated**: 2026-07-10

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README



# 知识图谱路径搜索系统

[English](#english-version) | [中文版本](#中文版本)

---

## 中文版本

### 项目简介

DaChuang 是一个基于 Neo4j 图数据库的知识图谱路径搜索与分析系统。该系统采用 DMCBA（Dynamic Multi-Criteria Backward Approaching）算法，能够在知识图谱中高效地搜索两个实体之间的最优路径，并提供实时进度反馈和可视化展示。

### 核心技术

- **后端框架**: FastAPI
- **图数据库**: Neo4j
- **异步任务**: Celery + Redis
- **实时通信**: WebSocket
- **算法**: DMCBA 动态多准则反向逼近算法

### 功能特性

1. **路径搜索**
   - 支持双向搜索（反向搜索 + 正向搜索）
   - 可配置权重参数（影响力、置信度、接近度）
   - 支持搜索深度和数量限制

2. **实时进度**
   - WebSocket 实时推送搜索进度
   - 显示各阶段完成状态

3. **图可视化**
   - 基于 G6 的交互式图谱展示
   - 节点/边的详细信息提示
   - 路径高亮显示

4. **AI 智能解释**
   - 基于用户意图的路径解释
   - 自动参数推荐

### 目录结构

```
DaChuang/
├── api/                    # API 路由模块
│   ├── agent.py           # AI Agent 接口
│   ├── fullgraph.py       # 全图获取接口
│   ├── search_task.py    # 搜索任务接口
│   └── websocket.py      # WebSocket 接口
├── services/             # 业务逻辑服务
│   ├── dmcba_service.py  # DMCBA 算法实现
│   └── dmcba_show.py  # 图展示服务
├── tasks/               # Celery 异步任务
│   └── path_search.py  # 路径搜索任务
├── db/                 # 数据库连接
│   ├── connection.py  # Neo4j 连接
│   └── graph.py     # 图配置
├── core/               # 核心组件
│   ├── celery_config.py     # Celery 配置
│   └── websocket_manager.py # WebSocket 管理
├── utils/              # 工具函数
│   └── caculator_tool.py # 计算器工具
├── DTO/               # 数据传输对象
│   └── shuju.py        # 数据模型
├── static/             # 前端静态资源
│   ├── graph.html     # 图谱可视化页面
│   └── g6.min.js    # G6 图库
└── main.py            # 应用入口
```

### 环境要求

- Python 3.10+
- Neo4j 数据库
- Redis 服务器

### 安装部署

1. **安装依赖**
```bash
pip install -r requirements.txt
```

2. **配置环境变量**
```bash
# 创建 .env 文件，配置以下变量：
NEO4J_URI=bolt://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your_password
REDIS_URL=redis://localhost:6379
```

3. **启动服务**
```bash
# 启动 Celery Worker
celery -A tasks worker --loglevel=info

# 启动 FastAPI 应用
uvicorn main:app --reload
```

4. **访问界面**
```
http://localhost:8000/graph
```

### API 接口

| 接口 | 方法 | 说明 |
|------|------|------|
| `/search/async` | POST | 发起异步路径搜索 |
| `/search/status/{task_id}` | GET | 获取任务状态 |
| `/search/result/{task_id}` | GET | 获取搜索结果 |
| `/fullgraph/async` | GET | 获取全图数据 |
| `/api/agent/interpret` | POST | AI 路径解释 |
| `/ws/progress/{task_id}` | WebSocket | 实时进度推送 |

### 算法参数说明

| 参数 | 默认值 | 说明 |
|------|--------|------|
| w1 | 0.3 | 影响力权重 |
| w2 | 0.3 | 置信度权重 |
| w3 | 0.4 | 接近度权重 |
| qoc_t | 0.2 | 时间质量阈值 |
| qoc_r | 0.4 | 可靠性阈值 |
| qoc_rho | 0.5 | 接近度阈值 |

### 许可证

本项目仅供学习研究使用。

---

## English Version

# Yelp Data-Based Social Path Intelligent Recommendation System

## Project Overview

This project is a **Provincial-level College Students’ Innovation and Entrepreneurship Development Program** (2025.05 - 2026.05). It is a precision social path recommendation system that quantifies social influence, trust value, and relationship closeness from Yelp user graphs. By storing user nodes and relationships in a Neo4j graph database, the system enables the rapid retrieval of high-value social paths. 

The project has obtained a **National Computer Software Copyright Registration (No. 2026SR0669313)**.

## Tech Stack

- **Frontend**: Vue3, AntV G6
- **Backend**: FastAPI (RESTful & WebSocket)
- **Graph Database**: Neo4j
- **Async Tasks**: Celery + Redis
- **Algorithm**: D-MCBA (Dynamic Multi-Criteria Backward Approaching) with Monte Carlo optimization
- **AI Model**: Qwen-Plus LLM

## Features

1. **Precision Social Path Search**
   - Quantifies social influence, trust value, and relationship closeness based on Yelp data.
   - Executes the D-MCBA algorithm with dynamic rewards and hop penalties.
   - Optimized search through Monte Carlo methods for rapid retrieval of high-value paths.

2. **Real-time Task Scheduling & Progress**
   - Integrated Celery with Redis for robust asynchronous task scheduling.
   - Built WebSocket interfaces using FastAPI for real-time progress streaming.

3. **Front-end Visualization**
   - Designed interactive front-end visualization pages with Vue3 and AntV G6.
   - Features node/edge tooltips and dynamic social path highlighting.

4. **AI Agent Empowerment**
   - Incorporated the Qwen-Plus LLM for user intent parsing.
   - Intelligent parameter tuning and automated search configuration.

## Directory Structure

```text
Yelp-Social-Path-System/
├── frontend/             # Vue3 Frontend (Vue3 + AntV G6)
│   ├── src/
│   │   ├── components/   # Graph visualization components
│   │   └── views/        # Main dashboard and interaction pages
├── api/                  # API routing modules (FastAPI)
│   ├── agent.py          # AI Agent interface (Qwen-Plus integration)
│   ├── fullgraph.py      # Full graph retrieval interface
│   ├── search_task.py    # Search task interface
│   └── websocket.py      # WebSocket interface
├── services/             # Business logic services
│   ├── dmcba_service.py  # D-MCBA algorithm & Monte Carlo implementation
│   └── dmcba_show.py     # Graph visualization data service
├── tasks/                # Celery async tasks
│   └── path_search.py    # Path search task
├── db/                   # Database connection
│   ├── connection.py     # Neo4j connection
│   └── graph.py          # Graph configuration
├── core/                 # Core components
│   ├── celery_config.py  # Celery configuration
│   └── websocket_manager.py # WebSocket manager
├── DTO/                  # Data transfer objects
│   └── shuju.py          # Data models
└── main.py               # Application entry point
### Quick Start

```bash
# Install dependencies
pip install -r requirements.txt

# Configure .env file
NEO4J_URI=bolt://localhost:7687
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your_password
REDIS_URL=redis://localhost:6379

# Start Celery Worker
celery -A tasks worker --loglevel=info

# Start FastAPI
uvicorn main:app --reload

# Access
http://localhost:8000/graph
```

### License

For educational and research purposes only.