Scale Decision Tree database

Summary

This repository contains a Docker image built over the standard Neo4j community edition container that pre-populates the graph database with the following CCS Guided Match decision tree structures:

• Linen & Laundry (RM6154)
• Wider public sector legal services (RM3788)
• Facilities Management 2 (Facilities)
• Facilities Management 2 (Housing)
• Facilities Management 2 (Security)
• Tech Products (RM3733), TePAS (RM6068) & EduTech (RM6103)

The structure for each decision tree is defined in Cypher Query Language and each script is executed against the graph database engine to load the data as the container is started.

Structure

The image is comprised of a standard Dockerfile which copies wrapper.sh and all resources from the graph subdirectory into the container. The wrapper.sh script (the entry point for the container) starts Neo4j via the standard docker-entrypoint.sh and waits until the service is stable before using the cypher-shell command line utility to connect and populate the graph using the Cypher scripts from the graph subdirectory. The script also sets the Neo4j admin account credentials and creates a service account based on the following environment variables:

• DB_ADMIN_USERNAME (default 'neo4j')
• DB_ADMIN_PASSWORD
• DB_SERVICE_ACCOUNT_USERNAME
• DB_SERVICE_ACCOUNT_PASSWORD

Build and Run

The Docker image can be built and tagged in a local development environment using:

docker build . -t scale/decision-tree-db:latest

To then start a container and expose the graph over the Neo4j Bolt protocol on port 7687, use:

docker run --rm -p 7687:7687 --env "DB_ADMIN_USERNAME=neo4j" --env "DB_ADMIN_PASSWORD=password" --env "DB_SERVICE_ACCOUNT_USERNAME=scale_dt" --env "DB_SERVICE_ACCOUNT_PASSWORD=password" scale/decision-tree-db:latest

Note we add the --rm flag to ensure the container is removed when stopped, and pass in values for environment variables that the wrapper.sh startup script expects.

Connect

The Neo4j browser can be used to connect to, query and visualise the graph: https://neo4j.com/developer/neo4j-browser/

With the container running on the local machine, connect via bolt://localhost:7687 using either the admin or service account credentials provided in the docker run command.

Example queries

Return all nodes in a single tree

MATCH (j:Journey {uuid: 'b87a0636-654e-11ea-bc55-0242ac130003'})-[fq:FIRST_QUESTION]->(fqi:QuestionInstance)
OPTIONAL MATCH p1 = (fqi)-[*]->(ag:AnswerGroup)-[:HAS_OUTCOME]->(:Agreement)-[:HAS_LOT]->(:Lot)
return j, fq, fqi, p1