Graph Data Modeling Example

Prerequsites

Neo4j Community Server Edition 3.4.x installed and started (https://neo4j.com/download-center/#community)
APOC-Awesome Procedures for Neo4j 3.4.x installed (https://github.com/neo4j-contrib/neo4j-apoc-procedures)
JDK 8 installed.

Repository structure

ACORD folder

ACORD.xml - ACORD Data Model 2.6

Scripts

load_xml00.txt - deletes any existing nodes and loads ACORD from xml. The XML file contains a lot of information but we only want to extract business concepts. In this example we only work with small subset and these nodes will be named POC1Type.

customStoredProcedures.txt - creates necessary stored procedures

logical01.txt - creates logical model. Creating logical model consists of series of steps, depending on what we wish to include.

physical02.txt -create physical model. We can now iter- ate, create, and recreate physical model based on the logical model.

transform03.txt- applies model transformations(e.g. splitting, merging, versioning etc) aka roll up and roll down

Schema/gql

Java program using Object Graph Mapper library to manage domain objects with Neo4j. Output is annotated schema file used by Amplify/Prisma in order to generate GrapQL API.

Setup

In <NEO4J_HOME>/conf/neo4j.conf add following properties

dbms.security.procedures.unrestricted=apoc.*
apoc.export.file.enabled=true
apoc.import.file.enabled=true
dbms.directories.plugins=plugins

Graph Data Modelling Steps

Load ACORD data model into Neo4j.

All you need to do to load XML files into the database is to call APOC procedure and provide url to the file call apoc.xml.import('file:///tmp/ACORD.xml');

Most convenient and performant way is to exectue these commands from cypher-shell as scripts. In this first step execute load_xml00.txt.

 ./cypher-shell -u neo4j -p <password> --format plain < /edh-model/scripts/load_xml00.txt

Create stored procedures

 ./cypher-shell -u neo4j -p <password> --format plain < /edh-model/scripts/customStoredProcedures.txt

Create logical layer.

Now let's execute logical01.txt

 ./cypher-shell -u neo4j -p <password> --format plain < /edh-model/scripts/logical01.txt

Create physicla layer

Execute physical02.txt

./cypher-shell -u neo4j -p <password> --format plain < /edh-model/scripts/physical02.txt

Model transformations

Let's apply transformations to the physicla model.

./cypher-shell -u neo4j -p <password> --format plain < /edh-model/scripts/transform03.txt

Data model can now be accessed through Neo4j browser.

Running the gqlschema

gqlschema.jar accepts command line arguments:

    Option (* = required)  Description                          
    ---------------------  -----------                          
   
    * --config             Path to the config file

Generate graphQL annotated schema

Run java program generating schema file.

java -jar gqlschema.jar --config conf/conf.yaml

Program Structure

You can read more about GQLSCHEMA here

conf - Configuration
images - Images
src - Java code
- graph - Graph traversals and application logic
- node - OGM classes
- util - utility code

Config

Config contains following parameters

server_uri - Uri of the Neo4j server.
server_username - username
server_password - password
schema - type of schema to be generated. "PRISMA" and "APPSYNC" are the valid options.

Example

Example of a conf.yaml file

server_uri: "bolt://localhost:7687"
server_username:  "neo4j"
server_password: "password123"
schema: "PRISMA"

afuyo/GraphQLSchema