Let LLM Think on Graph Schema Demo

This is a GenAI graph search Spring AI application written in Kotlin.

I use Customer 360 use case to prepare the sample graph data and schema.

Get started

The demo default use Ollama gemma2:9b model which minimally need ~6GB GPU VRAM to run.

Setup Neo4j graph database locally (docker compose)

In docker-compose.yml, there is a default NEO4J_AUTH environment variable.

docker-compose up -d

Inject sample graph data using Kotlin Notebook

Step 1: Initialize gradle project

Open IntelliJ IDEA and open build.gradle.kts and click sync. This will automatically download the project dependencies into your machine and be shared to Kotlin Notebook. Or run CLI

./gradlew build

Step 2: Install Kotlin notebook plugin in IDE

In IntelliJ IDEA, go to Settings > Plugins, Search Kotlin Notebook plugin. Install and restart.

Step 3: Run notebook

Open notebooks/neo4j.ipynb.

Click the play button to run notebook piece by piece. It should be able to connect and insert sample graph data.

Setup local Ollama or using cloud GenAI service on Spring AI

brew install ollama

ollama pull gemma2:9b

Run Spring Boot

Use intelliJ IDEA to run Spring Boot or using the following CLI.

./gradlew bootRun

Run Demo HTTP Request (app.http)

Open app.http, click the play button on the left to trigger http request to Spring Boot

Step by step to try the demo

Trace the output with prefix [DEMO

Change to cloud LLM provider from Ollama

If you computer doesn't have GPU or insufficient of GPU VRAM. You can use switch to other cloud LLMs provider that supported by Spring AI.

Step 1: Add other cloud LLM dependencies

Open build.gradle.kts, you can see the current dependency is using ollama org.springframework.ai:spring-ai-xxxxxxxxxx-spring-boot-starter

Update your preferred LLM provider.

Step 2: Modify model configuration

Open src/main/kotlin/configs/MoodelConfiguration, change the Ollama specific sources code to the new provider.

Open src/main/resources/application.properties, add configuration according to Spring AI documentation.