1 of 8

Neo4j and Generative AI

2 of 8

Pure Consumption Model

Neo4j Inc. All rights reserved 2024

  1. Siloed Approach
  2. Non-Connected Data
  3. No Fine-grained Access Control (RBAC)
  4. No Explainable AI using Graph Data Science

LLM

Structured

Unstructured

Ontologies

Data sources

Applications

Customer Service

Ticket Triaging

Recommendations

News Content & Discovery

Enterprise Knowledge Search

Patient Prioritization

Clinical Decision Support Systems

Pharmacovigilance

Health Assistants

FAQ Bots

3 of 8

Introducing Knowledge Graphs

Neo4j Inc. All rights reserved 2024

NEO4J KNOWLEDGE GRAPH

Why Neo4j

​

  • Connect related unstructured and structured data from multiple sources

​

  • Explainable AI & More predictability with Graph Data Science

​

  • Vector Embedding Support

Why Neo4j

​

  • Grounded Knowledge that is Contextual, Factual, Explainable and easy to trace the lineage

​

  • Grounded Facts, No Hallucinations!

​

  • Fine-grained Access Control (RBAC)

Ingesting Data into KG

Consuming Data from KG

Structured

Unstructured

Ontologies

Data sources

Applications

Customer Service

Ticket Triaging

Recommendations

News Content & Discovery

Enterprise Knowledge Search

Patient Prioritization

Clinical Decision Support Systems

Pharmacovigilance

Health Assistants

FAQ Bots

4 of 8

Neo4j and Generative AI Reference Architecture

Neo4j Inc. All rights reserved 2024

Knowledge graph

Graph Data Science

Graph DB

Applications

Parsing

Structured

Unstructured

Ontologies

Data sources

RAG

Customer Service

Ticket Triaging

Recommendations

News Content & Discovery

Enterprise Knowledge Search

Patient Prioritization

Clinical Decision Support Systems

Pharmacovigilance

Health Assistants

FAQ Bots

Bloom

APIs

VertexAI

with Generative AI

Neo4j Aura

APIs

VertexAI

with Generative AI

Ingesting Data into KG

Consuming Data from KG

5 of 8

Hands on Lab Architecture

Neo4j Inc. All rights reserved 2024

Input Data

Google Cloud Vertex AI

Workbench

Gradio Application

Graph Data Science

Graph Database

Bloom

Web Interface User

Generative AI

Neo4j AuraDS Professional

6 of 8

Lab 5 - Parsing

  • Zero-shot with a simple prompt with the LLM
  • Extract SEC EDGAR filing information in accordance with a Neo4j data model

Neo4j Inc. All rights reserved 2024

7 of 8

Lab 6 - Chatbot

  • Translates English to Cypher
  • Consumption using LLM with few shot prompting
  • Data augmentation from Neo4j response

Neo4j Inc. All rights reserved 2024

8 of 8

Lab 7 - Knowledge Graphs and Semantic Search

If your focus is analyzing documents on a file system, then vector indexing and search on text embeddings may be sufficient.

​

If you need to retrieve and make inferences about people, places, and things connected to those documents, knowledge graphs can help.

Neo4j Inc. All rights reserved 2024