KnowledgeGraphAI

project purpose and goals

• the code in this repository aims to showcase how
  • data from a legacy relational database can be transformed into a knowledge graph in Neo4j, and
  • enriched with AI capabilities for enhanced data retrieval and natural language processing

tools involved

• infrastructure
  • Docker: for MySQL, Neo4J containers
  • Google Cloud: for porting local application to the cloud
• Databases
  • MySQL: legacy relational database
  • Neo4j: graph database for knowledge graph creation
• Frontend
  • Streamlit: for building the frontend application
• AI
  • OpenAI/Gemini/Llama: large language models for query processing
  • LangChain: for integrating LLMs with the knowledge graph

sequence of steps

• this section lists the sequence of steps that will be taken as part of this project:
  • clone this repository locally and setup development environment
  • deploy MySQL and Neo4j containers using docker compose
  • load sample data into MySQL
  • move data from MySQL to Neo4J
  • build the knowledge graph inside Neo4J
  • develop frontend application with Streamlit
  • setup connection to LLM
  • implement query handling
  • port tool to cloud

installation and setup

• detailed steps on setting up the dev environment and other topics are available inside the setup folder

references

• Knowledge Graphs - A Practical Review of the Research Landscape
• Building Semantic Knowledge Graphs
• Knowledge Graphs - Research Paper
• O'Reilly - Building Knowledge Graphs Practitioner's Guide listing here and direct book here
• Neo4j Knowledge Graph Blog
• Knowledge Graph RAG Application
• Neo4j & LLM Fundamentals
• Introduction to Vector Indexes and Unstructured Data
• Building Neo4j Applications with Python
• LLM Graph Builder GitHub
• Neo4j LLM Knowledge Graph Builder
• LangChain Documentation
• Install Neo4J using Docker Compose