Kafka Link Prediction

This repo contains code that supports Confluent Blog post explaining how to use Neo4j together with Kafka.

The purpose of this code is to give a worked example of how to use the two together in a simple scenario, running inside of Docker. This example assumes that you have a working Confluent Cloud instance.

Setup

Setup the "fakestream" module for simulating data sent to Kafka:

• git submodule sync
• cd fakestream && pipenv install

Set up your topics in Kafka:

• Create a topic called "recommendations"
• Create a topic called "friends"

The "out of the box" configuration assumes you are using Confluent Cloud.

Export your configuration from confluent cloud and start the Neo4j server:

export CONFLUENT_API_KEY=(your value)
export CONFLUENT_API_SECRET=(your value)
export KAFKA_BOOTSTRAP_SERVERS=(your value)

docker-compose up

If you are not using Confluent Cloud and running Kafka locally

Inspect docker-compose.yml, uncomment the correct section, and put in your configuration.

Stream Data to Neo4j

Have a look at the run-stream.sh shell script for how to use fakestream to start a stream of friend requests.

Run the Examples

Take a look at the cypher files in this directory to follow along with examples from the blog post.