Graph Data Science (GDS) Overview & Projections

Graph Data Science (GDS) is a set of algorithms and tools that help analyze graph data in Neo4j. This lesson will provide an overview of GDS and focus on the projection features, which allow users to create in-memory graphs from existing data in the Neo4j database.

GDS Overview

The Graph Data Science Library allows for advanced graph analytics and machine learning on graph data. GDS provides algorithms for community detection, pathfinding, centrality, and more.

Key Concepts

• Graph Projection: Creating an in-memory representation of a subset of the graph for analysis.
• Algorithms: Predefined mathematical models to analyze graph structures.
• Data Models: The structure of the data and how it's represented in Neo4j.

Projections

Projections in GDS are essential for performing graph algorithms without querying the database each time. This is how you can create a projection:

Step-by-Step Process to Create a Projection

1. Define the Projection: Use the GDS projection methods to define nodes and relationships.
2. Run the Projection: Execute the projection command to create it in memory.
3. Execute Algorithms: Use the created projection to run graph algorithms.
4. Cleanup: Drop the projection when done to free up memory.

Example Code Snippet

CALL gds.graph.create(
    'myProjection',
    'Person',     // Node label
    'FRIENDS',    // Relationship type
    {
        relationshipProperties: 'weight'
    }
)
            

Best Practices

When working with GDS and projections, keep the following best practices in mind:

• Always drop projections after use to manage memory effectively.
• Use filtering options to limit the number of nodes and relationships in your projections.
• Test your algorithms with smaller projections before scaling up.

FAQ

CALL gds.graph.drop('myProjection')