Skip to content
  • https://github.com/vesoft-inc/nebula
  • https://www.galaxybase.com/
  • https://www.tigergraph.com/

  • https://docs.aws.amazon.com/neptune/latest/userguide/intro.html

  • Memgraph 主打在数据流上做图计算

  • Katana Graph 图数据库与AI相结合
  • ArangoDB 多模态数据库

Applications

Batch

  • Identifying Essential Proteins with Betweenness Centrality in PPI network

Streaming

  • Real-time Reddit analysis

Both Batch & Streaming

  • Fraud Detection
  • Recommendation Engine

Applications (Examples)

Life Science

  • Graph Queries (e.g., virtual screening of ChEMBL database, trial management)
  • Graph Analytics (substructure searches, similarity search, fingerprint generation)
  • Graph Mining (pattern discovery)
  • Graph AI & Deep Machine Learning (patient classification, drug retargeting, precision medicine)

Chart: Life Sciences 2

  • Financial Services
    • contextual search (Graph Queries)
    • path finding, centrality and community detection (Graph Analytics)
    • pattern discovery (Graph Mining)
    • deep learning and prediction (Graph AI)
  • Life Science
    • virtual screening of ChEMBL database, trial management (Graph Queries)
    • substructure searches, similarity search, fingerprint generation (Graph Analytics)
    • pattern discovery (Graph Mining)
    • patient classification, drug retargeting, precision medicine (Graph AI)

Chart: Financial Overview 1

GDB Benchmark

https://github.com/galaxybase/graph-database-benchmark

https://github.com/RedisGraph/graph-database-benchmark

数据加载测试

  1. 加载耗时
  2. 加载后数据占用的磁盘容量

数据查询测试

  1. K-hop neighbor 响应时间
  2. Shortest Path 响应时间
  3. PageRank 响应时间
  4. Weakly Connected Components 响应时间
  5. Label Propagation Algorithm 响应时间
  6. Degree Centrality 响应时间

Graph Platform Applications

Overview

Summary: Most applications are built by directly executing several graph queries (e.g., the recommendation engine), and graph analytics applications (e.g., centrality computation) are powered by an algorithm package built on their GDB.

Basically, the applications can be roughly categorized into

  • Graph Queries
  • key performance: k-hop neighborhood latency
  • Graph Analytics
  • provided by an algorithm library built on their GDB
  • key performance: graph algorithm latency
    • Shortest Path
    • PageRank
    • Weakly Connected Components
    • Label Propagation Algorithm
    • Degree Centrality
  • Graph Mining
    • pattern matching/discovery
  • Graph AI
  • no explicit benchmark yet

And actually, the applications of different companies are usually not different. Some common applications are summarized below:

  • Recommendation Engine
  • 深度关联推荐 (Graph Query -> Multi-hop (3-10 hops) Query, Graph Analytics -> centrality/similarity)
  • 实时推荐引擎 (Graph Query -> Low Query Latency)
  • 人工智能和机器学习构建新一代推荐引擎 (Graph AI)
  • Fraud Detection (Graph Query, Graph Analytics -> pattern matching/discovery, Graph AI)
  • Life Science
  • virtual screening of ChEMBL database, trial management (Graph Queries)
  • substructure searches, similarity search, fingerprint generation (Graph Analytics)
  • pattern discovery (Graph Analytics)
  • patient classification, drug retargeting, precision medicine (Graph AI)

  • Knowledge Graph (Knowledge Graph Query -> KG query/reasoning, Graph Analytics -> pattern discovery, Graph AI -> KG Embedding)

  • 影响力营销

  • 深度关联分析 (Graph Query, Graph Analytics -> Similarity/Pattern Matching)
  • 找到最有影响力的人 (Graph Analytics -> PageRank)
  • 挖掘KOL覆盖的社交圈 (Graph Analytics -> Community Detection)

Neo4j

Features

Single-machine;

Products

  • Neo4j Graph Database
  • A highly scalable, native graph database, purpose built to persist and protect relationships.
  • Neo4j Graph Data Science™ Library
  • A toolkit with a flexible data structure for analytics and a library with six varieties of powerful graph algorithms.
  • Algorithms
    • Community Detection
    • Centrality (Importance)
    • Similarity
    • Heuristic Link Prediction
    • Pathfinding & Search
    • Node Embedding
  • Neo4j Bloom
  • A graph visualization and exploration tool that allows users to visualize algorithm results and find patterns using codeless search.

Applications

  • Real-Time Recommendations (Graph Query, Graph Analytics -> centrality)

  • supports many named, directed relationships between entities (or nodes) which give a rich semantic context for the data

  • Fraud Detection (Graph Query, Graph Analytics -> pattern matching/discovery, Graph AI)

  • first-party bank fraud
  • credit card fraud
  • ecommerce fraud
  • insurance fraud
  • money laundering
  • Knowledge Graph (Knowledge Graph Query -> KG query/reasoning, Graph Analytics -> pattern discovery, Graph AI -> KG Embedding)
  • Actioning Knowledge Graph (to drive action)
    • data assurance
    • data discovery
    • data insight
  • Decisioning Knowledge Graph (to improve decisions)
    • graph analytics
    • machine learning
    • data science
  • Life Sciences
  • Drug Discovery (Graph Analytics, Graph AI)
  • Medical Knowledge Graph (Knowledge Graph Query, Graph Analytics, Graph AI)
  • dealing with everything from patients to molecules
    • understand the value of graphs for R&D, privacy and regulatory compliance, medical equipment manufacturing and affiliation management between healthcare providers (HCPs), patients and organizations
  • Social Networking (Graph Query)
  • Social media networks are already graphs, so there's no point converting a graph into tables and then back again.

TigerGraph

Features

Single-machine;

Supported graph algorithms:

https://github.com/tigergraph/gsql-graph-algorithms

Applications

  • 个性化推荐引擎
  • 深度关联推荐 (Graph Query -> Multi-hop (3-10 hops) Query)
  • 实时推荐引擎 (Graph Query -> Low Query Latency)
  • 人工智能和机器学习构建新一代推荐引擎 (Graph AI)
  • 影响力营销
  • 深度关联分析 (Graph Query, Graph Analytics -> Similarity/Pattern Matching)
  • 找到最有影响力的人 (Graph Analytics -> PageRank)
  • 挖掘KOL覆盖的社交圈 (Graph Analytics -> Community Detection)
  • 反洗钱
  • 实时深度链路分析 (Graph Query)
  • 深度关联分析 (Graph Query, Graph Analytics -> Similarity/Pattern Matching)
  • 机器学习 (Graph AI)
  • 供应链优化
  • 深度关联分析 (Graph Query -> Multi-hop Query, Graph Analytics -> Pattern Discovery)
  • 实时分析 (Visualization)
  • 机器学习 (Graph AI)

NebulaGraph

Features

Distributed;

Applications

  • 知识图谱 (Knowledge Graph Query -> KG query/reasoning, Graph Analytics -> pattern discovery, Graph AI -> KG Embedding)
  • 安全风控 (Graph Query -> Multi-hop Query, Graph Analytics -> Pattern Discovery)
  • 链路分析 (Graph Analytics -> Pattern Discovery)
  • 组织架构管理 (Graph Query -> Multi-hop Query)

KanataGraph

Features

Distributed; All-in-one graph platform;

Applications

  • Financial Services
  • contextual search (Graph Queries)
  • path finding, centrality and community detection (Graph Analytics)
  • pattern discovery (Graph Analytics)

  • deep learning and prediction (Graph AI)

  • Life Science

  • virtual screening of ChEMBL database, trial management (Graph Queries)
  • substructure searches, similarity search, fingerprint generation (Graph Analytics)
  • pattern discovery (Graph Analytics)
  • patient classification, drug retargeting, precision medicine (Graph AI)

Memgraph

Features

Real-time/Streaming graph analytics;

Applications

  • Fraud Detection (Graph Query, Graph Analytics)
  • Community Detection (Graph Analytics)
  • Online Dynamic Community Detection
  • PageRank (Graph Analytics -> PageRank)
  • Dynamic PageRank
  • Identify Essential Proteins (Graph Analytics -> Betweenness Centrality)
  • Route Computation (Graph Query)
  • Recommendation Engine (Graph Query)

ArangoDB

Features

Multi-model database; focusing graph analytics;

Applications

  • Enterprise Knowledge Graphs
  • harmonizing internal and external data relevant to an organization into a common semantic model
  • Adaptive Fraud Detection & Analytics
  • Single View of Everything
  • scalable and adaptive