Entry G11: Create the Marvel Multigraph Database

Creating a multigraph database was actually way easier than I expected.

I’m still using the Marvel Universe Social Network and the directions for creating the initial database are exactly the same as the directions from Entry G2. For easy reference, I’ll include most the steps, directions, and pictures, but some may be more condensed. I’m also assuming you’ve got the Neo4j Desktop downloaded (if not, see Entry G2)

Create an Empty Database

1. Click the Add button in the upper right of the My Project area

2. Choose Local DBMS

3. Give your database a name and set a password and version

4. Admire your new database

Setup the database

Just like before, we need to prepare the database for data.

1. Click on your new database

This opens the Options for the database. It defaults to the Details page.

2. Navigate to Plugins

We want to add some plugins to make working with the data easier. To do this, choose the Plugins tab. You’ll see the four available plugins

3. Install the desired plugins

1. Expland the APOC and Graph Data Science libraries individually
2. Click the Install button for each

Fire up the database

1. Start the database

Click Start to start the database

2. Open the database

Click Open to fire up the browser based interface

You’ll notice that once the database is running, the options change slightly; Start has changed to Stop and Open is now selectable.

3. Navigate to Neo4j Browser

You should now have a Neo4j Desktop Browser window open. The initial page usually looks something like this (I’m using Neo4j Desktop version 1.4.1 and database version 4.1.3)

Create Multiple Graphs

This is where things start to change. The code in the G11 notebook starts here.

The Managing Multiple Databases in Neo4j tutorial in the Neo4j Developer Guides has really nice, easy to follow directions on how to create multigraph databases in Neo4j. Here are the directions on how to use those directions for our purposes.

1. Access the system database

A Neo4j database is created with two default databases:

• the default neo4j database
• a system database

The default neo4j database is the one we’ve been using. It’s the standard database that holds graph data.

The system database holds information about the databases in the Neo4j instance. It is also where we create/edit/delete graphs within the instance.

To access this database type :use system into the Neo4j command line:

You can type show databases to see the two existing databases:

2. Create a database for each model

Creating a new, empty database is really easy, just type create database db_name into the Neo4j command line. Of course, you’ll need to specify the actual name of the database:

Populate the Bimodal Graph

1. Go to the Bimodal Graph

This is much easier than Starting and Stopping graphs in the Project area of Neo4j Desktop. Just enter the following into the Neo4j command line:

:use db_name

If you used the same naming convention I did, then it will look like this:

You are now in the bimodal graph.

2. Create the schema

Now we can use the exact same instructions from Entry G2 to load the data.

Input the following in the Neo4j command line to set the schema:

CALL apoc.schema.assert( {},
{Comic:['name'],Hero:['name']});

3. Load the data

Pull the data directly from Tomaz’s github page into the graph:

CALL apoc.load.csv('https://raw.githubusercontent.com/tomasonjo/neo4j-marvel/master/data/edges.csv') yield map as row WITH row
MERGE (h:Hero {name:row.hero})
MERGE (c:Comic {name:row.comic})
MERGE (h)-[:APPEARS_IN]->(c);

Populate the Mixed Graph

This is even easier than cloning the graph like we did in Entry G5.

Let’s condense the directions from populating the bimodal graph above and Entry G5:

1. Enable Multiple Statements

Mine was on by default, but make sure that “Enable multi statement query editor” is checked in the Browser Settings (the gear icon):

2. Go to the Mixed Graph

If you used the same naming convention I did, the code will look like this:

:use mixmodal

3. Set schema, load data, and project unimodal model

We can combine our statements and let Neo4j run them all one after another:

CALL apoc.schema.assert( {},
{Comic:['name'],Hero:['name']});

CALL apoc.load.csv('https://raw.githubusercontent.com/tomasonjo/neo4j-marvel/master/data/edges.csv') yield map as row WITH row
MERGE (h:Hero {name:row.hero})
MERGE (c:Comic {name:row.comic})
MERGE (h)-[:APPEARS_IN]->(c);

CALL apoc.periodic.iterate('MATCH (h1:Hero)-->(:Comic)<--(h2:Hero) where id(h1) < id(h2) RETURN h1, h2',
'MERGE (h1)-[r:KNOWS]-(h2) on CREATE SET r.weight = 1 on MATCH SET r.weight = r.weight+1', {batchSize:5000, parallel:false, iterateList:True});

Side note, I put an empty line between the different statements to make it clear what code belongs to which statement. However, there is no need for this.

Populate the Unimodal Graph

Last but not least, we create our unimodal model graph.

1. Go to the Unimodal Graph

If you used the same naming convention I did, the code will look like this:

:use unimodal

3. Set schema, load data, project unimodal model, and delete bimodal elements

We use the same statements as for the mixed model, then add a statement to remove the Comic nodes and their relationships:

CALL apoc.schema.assert( {},
{Comic:['name'],Hero:['name']});

CALL apoc.load.csv('https://raw.githubusercontent.com/tomasonjo/neo4j-marvel/master/data/edges.csv') yield map as row WITH row
MERGE (h:Hero {name:row.hero})
MERGE (c:Comic {name:row.comic})
MERGE (h)-[:APPEARS_IN]->(c);

CALL apoc.periodic.iterate('MATCH (h1:Hero)-->(:Comic)<--(h2:Hero) where id(h1) < id(h2) RETURN h1, h2',
'MERGE (h1)-[r:KNOWS]-(h2) on CREATE SET r.weight = 1 on MATCH SET r.weight = r.weight+1', {batchSize:5000, parallel:false, iterateList:True});

MATCH (c:Comic)
DETACH DELETE c;

Check results

We can now see all of our databases have been created by switching back to the system database (to see the results of show database you will need to run these lines separately, not as multiple statements).

:use system

show databases

That’s it for creating the databases. So far I’m finding this much nicer than having each model in its own separate instance. I can now easily switch between graph models while remaining in a Jupyter notebook. This will allow me to group topics much more logically instead of running the code for several topics in a single notebook and having one for each graph model.

Up Next

Degree Comparison

Resources

• Marvel Universe Social Network
• Graph People blog
• Tomaz Bratanic’s github page
• Managing Multiple Databases in Neo4j
• Neo4j Developer Guides
• Entry G2: Create a Neo4j Database
• Entry G5: Projecting Bimodal to Unimodal