Skip to content
Merged
Show file tree
Hide file tree
Changes from 2 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion content/cn/blog/_index.md
Original file line number Diff line number Diff line change
Expand Up @@ -14,4 +14,4 @@ menu:
- **HugeGraph-AI**: AI 与图计算结合的应用案例
- **HugeGraph Computing (OLAP)**: 大规模图计算和分析实践

探索更多内容,提升您的图数据库使用体验
探索更多内容,深入了解 HugeGraph 图系统的全栈能力
2 changes: 1 addition & 1 deletion content/cn/docs/guides/architectural.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ weight: 1

### 1 概述

作为一款通用的图数据库产品,HugeGraph 需具备图数据库的基本功能。HugeGraph 支持 OLTP 和 OLAP 两种图计算类型其中 OLTP 实现了 [Apache TinkerPop3](https://tinkerpop.apache.org) 框架,支持 [Gremlin](https://tinkerpop.apache.org/gremlin.html) 和 [Cypher](https://en.wikipedia.org/wiki/Cypher) 查询语言,拥有功能齐全的应用工具链,还提供了插件式后端存储驱动框架。
作为一套覆盖**图数据库、图计算与图 AI** 的全栈图系统,HugeGraph 以高性能图引擎(HugeGraph Server)为核心,支持 OLTP 和 OLAP 两种图计算类型其中 OLTP 层实现了 [Apache TinkerPop3](https://tinkerpop.apache.org) 框架,支持 [Gremlin](https://tinkerpop.apache.org/gremlin.html) 和 [Cypher](https://en.wikipedia.org/wiki/Cypher) 查询语言,拥有功能齐全的应用工具链,还提供了插件式后端存储驱动框架。

下面是 HugeGraph 的整体架构图:

Expand Down
201 changes: 124 additions & 77 deletions content/cn/docs/introduction/_index.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,95 +7,142 @@ aliases:
- /cn/docs/introduction/README/
---

### Summary
### 什么是 Apache HugeGraph?

Apache HugeGraph 是一款易用、高效、通用的开源图数据库系统(Graph Database,[GitHub 项目地址](https://github.com/apache/hugegraph)),
实现了[Apache TinkerPop3](https://tinkerpop.apache.org)框架及完全兼容[Gremlin](https://tinkerpop.apache.org/gremlin.html)查询语言,
同时支持 [Cypher](https://opencypher.org/) 查询语言(OpenCypher 标准),
具备完善的工具链组件,助力用户轻松构建基于图数据库之上的应用和产品。HugeGraph 支持百亿以上的顶点和边快速导入,并提供毫秒级的关联关系查询能力(OLTP),
并支持大规模分布式图分析(OLAP)。
[Apache HugeGraph](https://hugegraph.apache.org/) 是一套易用、高效、通用的开源**全栈图系统**([GitHub](https://github.com/apache/hugegraph)),
覆盖**图数据库**(OLTP 实时查询)、**图计算**(OLAP 大规模分析)与**图 AI**(GraphRAG / 图机器学习)三大领域。

HugeGraph 典型应用场景包括深度关系探索、关联分析、路径搜索、特征抽取、数据聚类、社区检测、知识图谱等,
适用业务领域有如网络安全、电信诈骗、金融风控、广告推荐、社交网络和智能机器人等。
HugeGraph 支持百亿以上的顶点和边的快速存储与查询,具备出色的 OLTP 性能。
其图引擎完全兼容 [Apache TinkerPop 3](https://tinkerpop.apache.org) 框架,同时支持
[Gremlin](https://tinkerpop.apache.org/gremlin.html) 和 [Cypher](https://en.wikipedia.org/wiki/Cypher)(OpenCypher 标准)双查询语言。

本系统的主要应用场景是解决反欺诈、威胁情报、黑产打击等业务的图数据存储和建模分析需求,在此基础上逐步扩展及支持了更多的通用图应用。
**典型应用场景:** 深度关系探索、关联分析、路径搜索、特征抽取、社区检测、知识图谱等,
**适用领域:** 网络安全、电信反欺诈、金融风控、广告推荐、社交网络、智能问答等。

### Features
---

### 生态系统全景

```
┌──────────────────────────────────────────────────────────────┐
│ Apache HugeGraph - Full-Stack Graph System │
├──────────────────┬────────────────────┬──────────────────────┤
│ Graph DB (OLTP) │ Graph Compute │ Graph AI │
│ HugeGraph │ Vermeer (Memory) │ HugeGraph-AI │
│ Server │ Computer (Dist.) │ GraphRAG/GNN/Py │
├──────────────────┴────────────────────┴──────────────────────┤
│ HugeGraph Toolchain │
│ Hubble | Loader | Client(Java/Go/Py) | Spark | Tools │
└──────────────────────────────────────────────────────────────┘
```

---

### 核心组件

#### 🗄️ HugeGraph Server — 图引擎(OLTP)

HugeGraph 项目的核心模块,提供高性能的图数据存储与实时查询能力:

- **图引擎核心**:支持属性图(Property Graph)建模,包含 VertexLabel、EdgeLabel、PropertyKey、IndexLabel 完整 Schema 管理
- **双查询语言**:全面兼容 Gremlin(TinkerPop 3)和 Cypher(OpenCypher)
- **REST API**:内置 REST Server,提供 RESTful 图操作接口
- **多类型索引**:精确查询、范围查询、复合条件组合查询
- **插件式存储后端**:1.7.0+ 默认支持 `RocksDB`(单机默认)、`HStore`(分布式)、`HBase`、`Memory`,1.5.x 及以前还支持 MySQL / PostgreSQL / Cassandra 等

**子模块:**
- `Core` — 图引擎实现,向下连接 Backend,向上支持 API
- `Backend` — 多后端存储适配层
- `API` — RESTful 接入层,兼容 Gremlin/Cypher 查询

📖 [Server 快速开始](/cn/docs/quickstart/hugegraph/hugegraph-server)

---

#### 📊 图计算引擎(OLAP)

提供两种互补的图分析引擎:

- **Vermeer**(推荐):高性能纯内存图计算引擎,部署简单、响应快,适合中小规模图分析和快速上手
- **HugeGraph-Computer**:基于 [Pregel](https://kowshik.github.io/JPregel/pregel_paper.pdf) 的分布式 OLAP 引擎,可运行在 Kubernetes / Yarn 上,适合超大规模图算法任务

📖 [图计算快速开始](/cn/docs/quickstart/computing/hugegraph-computer)

---

#### 🤖 HugeGraph-AI — 图 AI 生态

HugeGraph 支持在线及离线环境下的图操作,支持批量导入数据,支持高效的复杂关联关系分析,并且能够与大数据平台无缝集成。
HugeGraph 支持多用户并行操作,用户可输入 Gremlin/Cypher 查询语句,并及时得到图查询结果,也可在用户程序中调用 HugeGraph API 进行图分析或查询。
HugeGraph 独立的 AI 组件,连接图与大语言模型(LLM):

本系统具备如下特点:
- **GraphRAG**:基于图的检索增强生成,实现 LLM 智能问答
- **知识图谱构建**:自动从非结构化文本中提取实体和关系,构建知识图谱
- **图神经网络**:支持 GNN 模型的训练与推理
- **20+ 图机器学习算法**:内置丰富的图分析算法,持续更新
- **Python Client**:为 AI 应用提供便捷的 Python SDK

- 易用:HugeGraph 支持 Gremlin/Cypher 图查询语言与 RESTful API,同时提供图检索常用接口,具备功能齐全的周边工具,轻松实现基于图的各种查询分析运算。
- 高效:HugeGraph 在图存储和图计算方面做了深度优化,提供多种批量导入工具,轻松完成百亿级数据快速导入,通过优化过的查询达到图检索的毫秒级响应。支持数千用户并发的在线实时操作。
- 通用:HugeGraph 支持 Apache Gremlin 标准图查询语言和 Property Graph 标准图建模方法,支持基于图的 OLTP 和 OLAP 方案。集成 Apache Hadoop 及 Apache Spark 大数据平台。
- 可扩展:支持分布式存储、数据多副本及横向扩容,内置多种后端存储引擎,也可插件式轻松扩展后端存储引擎。
- 开放:HugeGraph 代码开源(Apache 2 License),客户可自主修改定制,选择性回馈开源社区。
📖 [HugeGraph-AI 快速开始](/cn/docs/quickstart/hugegraph-ai/quick_start)

---

#### 🛠️ HugeGraph Toolchain — 工具链

围绕图系统的完整工具生态([toolchain 仓库](https://github.com/apache/hugegraph-toolchain)):

| 工具 | 说明 |
|------|------|
| [Hubble](/cn/docs/quickstart/toolchain/hugegraph-hubble) | Web 可视化平台:数据建模 → 批量导入 → 在线/离线分析 一站式操作 |
| [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) | 数据导入工具:支持本地文件、HDFS、MySQL 等多数据源,TXT/CSV/JSON 等格式 |
| [Client](/cn/docs/quickstart/client/hugegraph-client) | 多语言 SDK:Java / Python / Go |
| [Spark-connector](/cn/docs/quickstart/toolchain/hugegraph-spark-connector) | Spark 集成:支持通过 Spark 批量读写图数据,适合大数据离线处理场景 |
| [Tools](/cn/docs/quickstart/toolchain/hugegraph-tools) | 命令行运维工具:图管理、备份恢复、Gremlin 执行等 |

---

### 部署模式

HugeGraph 支持多种部署模式,满足不同规模和场景的需求:
HugeGraph 支持两种主要部署模式:

| 模式 | 核心组件 | 适用场景 | 数据规模 | 高可用 |
|------|---------|---------|---------|-------|
| **单机 (Standalone)** | Server + RocksDB | 开发、测试、单节点生产 | < 4TB | 基础 |
| **分布式 (Distributed)** | Server + PD(3-5节点)+ Store(3+节点) | 生产环境、水平扩展 | < 1000TB | ✅ |

**单机模式 (Standalone)**
- Server + RocksDB 后端存储
- 适合开发测试和中小规模数据(< 4TB)
- Docker 快速启动: `docker run hugegraph/hugegraph`
- 详见 [Server 快速开始](/cn/docs/quickstart/hugegraph/hugegraph-server)
**Docker 快速体验:**

**分布式模式 (Distributed)**
- HugeGraph-PD: 元数据管理和集群调度
- HugeGraph-Store (HStore): 分布式存储引擎
- 支持水平扩展和高可用(< 1000TB 数据规模)
- 适合生产环境和大规模图数据应用
```bash
docker run -itd --name=hugegraph -p 8080:8080 hugegraph/hugegraph
```

---

### 快速入门指南
### 快速入门导航

| 使用场景 | 推荐路径 |
| 我想要... | 从这里开始 |
|---------|---------|
| 快速体验 | [Docker 部署](/cn/docs/quickstart/hugegraph/hugegraph-server#docker) |
| 构建 OLTP 应用 | Server → REST API / Gremlin / Cypher |
| 图分析 (OLAP) | [Vermeer](/cn/docs/quickstart/computing/hugegraph-computer) (推荐) 或 Computer |
| 构建 AI 应用 | [HugeGraph-AI](/cn/docs/quickstart/hugegraph-ai) (GraphRAG/知识图谱) |
| 批量导入数据 | [Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) + [Hubble](/cn/docs/quickstart/toolchain/hugegraph-hubble) |

### 功能特性

- 支持从多数据源批量导入数据 (包括本地文件、HDFS 文件、MySQL 数据库等数据源),支持多种文件格式导入 (包括 TXT、CSV、JSON 等格式)
- 具备可视化操作界面,可用于操作、分析及展示图,降低用户使用门槛
- 优化的图接口:最短路径 (Shortest Path)、K 步连通子图 (K-neighbor)、K 步到达邻接点 (K-out)、个性化推荐算法 PersonalRank 等
- 基于 Apache TinkerPop3 框架实现,支持 Gremlin 图查询语言
- 支持属性图,顶点和边均可添加属性,支持丰富的属性类型
- 具备独立的 Schema 元数据信息,拥有强大的图建模能力,方便第三方系统集成
- 支持多顶点 ID 策略:支持主键 ID、支持自动生成 ID、支持用户自定义字符串 ID、支持用户自定义数字 ID
- 可以对边和顶点的属性建立索引,支持精确查询、范围查询、全文检索
- 存储系统采用插件方式,支持 RocksDB(单机/集群)、Cassandra、ScyllaDB、HBase、MySQL、PostgreSQL、Palo 以及 Memory 等
- 与 HDFS、Spark/Flink、GraphX 等大数据系统集成,支持 BulkLoad 操作导入海量数据
- 支持高可用 HA、数据多副本、备份恢复、监控、分布式 Trace 等

### Modules

- [HugeGraph-Server](/cn/docs/quickstart/hugegraph/hugegraph-server): HugeGraph-Server 是 HugeGraph 项目的核心部分,包含 Core、Backend、API 等子模块;
- Core:图引擎实现,向下连接 Backend 模块,向上支持 API 模块;
- Backend:实现将图数据存储到后端,支持的后端包括:Memory、Cassandra、ScyllaDB、RocksDB、HBase、MySQL 及 PostgreSQL,用户根据实际情况选择一种即可;
- API:内置 REST Server,向用户提供 RESTful API,同时完全兼容 Gremlin 查询。(支持分布式存储和计算下推)
- [HugeGraph-Toolchain](https://github.com/apache/hugegraph-toolchain): (工具链)
- [HugeGraph-Client](/cn/docs/quickstart/client/hugegraph-client):HugeGraph-Client 提供了 RESTful API 的客户端,用于连接 HugeGraph-Server,支持 Java/Python/Go 多语言版本;
- [HugeGraph-Loader](/cn/docs/quickstart/toolchain/hugegraph-loader):HugeGraph-Loader 是基于 HugeGraph-Client 的数据导入工具,将普通文本数据转化为图形的顶点和边并插入图形数据库中;
- [HugeGraph-Hubble](/cn/docs/quickstart/toolchain/hugegraph-hubble):HugeGraph-Hubble 是 HugeGraph 的 Web
可视化管理平台,一站式可视化分析平台,平台涵盖了从数据建模,到数据快速导入,再到数据的在线、离线分析、以及图的统一管理的全过程;
- [HugeGraph-Tools](/cn/docs/quickstart/toolchain/hugegraph-tools):HugeGraph-Tools 是 HugeGraph 的部署和管理工具,包括管理图、备份/恢复、Gremlin 执行等功能。
- [HugeGraph-Computer](/cn/docs/quickstart/computing/hugegraph-computer):HugeGraph-Computer 是分布式图处理系统 (OLAP)。
它是 [Pregel](https://kowshik.github.io/JPregel/pregel_paper.pdf) 的一个实现。它可以运行在 Kubernetes/Yarn
等集群上,支持超大规模图计算。同时提供 Vermeer 轻量级图计算引擎,适合快速开始和中小规模图分析。
- [HugeGraph-AI](/cn/docs/quickstart/hugegraph-ai):HugeGraph-AI 是 HugeGraph 独立的 AI
组件,提供 LLM/GraphRAG 智能问答、自动化知识图谱构建、图神经网络训练/推理、Python-Client 等功能,内置 20+ 图机器学习算法,持续更新中。

### Contact Us

- [GitHub Issues](https://github.com/apache/hugegraph/issues): 使用途中出现问题或提供功能性建议,可通过此反馈 (推荐)
- 邮件反馈:[dev@hugegraph.apache.org](mailto:dev@hugegraph.apache.org) ([邮箱订阅方式](https://hugegraph.apache.org/docs/contribution-guidelines/subscribe/))
- SEC 反馈: [security@hugegraph.apache.org](mailto:security@hugegraph.apache.org) (报告安全相关问题)
- 微信公众号:Apache HugeGraph, 欢迎扫描下方二维码加入我们!

<img src="https://github.com/apache/hugegraph-doc/blob/master/assets/images/wechat.png?raw=true" alt="QR png" width="300"/>
| 🚀 **快速体验** | [Docker 部署](/cn/docs/quickstart/hugegraph/hugegraph-server) |
| 🔍 **运行图查询** (OLTP) | [HugeGraph Server 快速开始](/cn/docs/quickstart/hugegraph/hugegraph-server) |
| 📈 **大规模图计算** (OLAP) | [Vermeer / Computer](/cn/docs/quickstart/computing/hugegraph-computer) |
| 🤖 **构建 AI/RAG 应用** | [HugeGraph-AI](/cn/docs/quickstart/hugegraph-ai/quick_start) |
| 📥 **批量导入数据** | [HugeGraph Loader](/cn/docs/quickstart/toolchain/hugegraph-loader) |
| 🖥️ **可视化管理** | [Hubble Web UI](/cn/docs/quickstart/toolchain/hugegraph-hubble) |

---

### 系统特性

- **易用**:Gremlin/Cypher 双查询语言 + RESTful API,功能齐全的工具链,轻松上手
- **高效**:图存储与查询深度优化,毫秒级响应,支持数千并发在线操作,百亿级数据快速导入
- **通用**:支持 OLTP + OLAP 双模式,无缝对接 Apache Hadoop、Spark、Flink 大数据生态
- **可扩展**:分布式存储、数据多副本、横向扩容,插件式后端可灵活扩展
- **开放**:Apache 2.0 License,完全开源,欢迎社区贡献

---

### 联系我们

- [GitHub Issues](https://github.com/apache/hugegraph/issues):问题反馈与功能建议(推荐)
- 邮件:[dev@hugegraph.apache.org](mailto:dev@hugegraph.apache.org)([订阅方式](/cn/docs/contribution-guidelines/subscribe/))
- 安全问题:[security@hugegraph.apache.org](mailto:security@hugegraph.apache.org)
- 微信公众号:Apache HugeGraph

<img src="https://github.com/apache/hugegraph-doc/blob/master/assets/images/wechat.png?raw=true" alt="微信公众号二维码" width="300"/>
2 changes: 1 addition & 1 deletion content/en/blog/_index.md
Original file line number Diff line number Diff line change
Expand Up @@ -14,4 +14,4 @@ Here you will find technical articles, best practices, and community updates fro
- **HugeGraph-AI**: Application cases combining AI with graph computing
- **HugeGraph Computing (OLAP)**: Large-scale graph computing and analytical practices

Explore more content to enhance your graph database experience!
Explore more content to deeply understand the full-stack capabilities of the HugeGraph system!
2 changes: 1 addition & 1 deletion content/en/docs/guides/architectural.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ weight: 1

### 1 Overview

As a general-purpose graph database product, HugeGraph needs to possess basic graph database functionality. HugeGraph supports two types of graph computation: OLTP and OLAP. For OLTP, it implements the [Apache TinkerPop3](https://tinkerpop.apache.org) framework and supports the [Gremlin](https://tinkerpop.apache.org/gremlin.html) and [Cypher](https://en.wikipedia.org/wiki/Cypher) query languages. It comes with a complete application toolchain and provides a plugin-based backend storage driver framework.
As a full-stack graph system covering **Graph Database, Graph Computing, and Graph AI**, HugeGraph is centered around a high-performance graph engine (HugeGraph Server) and supports both OLTP and OLAP graph computation types. For the OLTP layer, it implements the [Apache TinkerPop3](https://tinkerpop.apache.org) framework and supports the [Gremlin](https://tinkerpop.apache.org/gremlin.html) and [Cypher](https://en.wikipedia.org/wiki/Cypher) query languages. It comes with a complete application toolchain and provides a pluggable backend storage driver framework.

Below is the overall architecture diagram of HugeGraph:

Expand Down
Loading