Day 21: Dgraph - Graph Data Modeling for Knowledge Graphs

​

What you’ll accomplish today

• Understand key Dgraph concepts and architecture
• Load sample RDF data from a news article knowledge graph
• Explore different types of knowledge graphs (lexical, domain, visual, geospatial)
• Model complex relationships and schema design
• Connect to the hyper-news example project for real-world data

​

Understanding key Dgraph concepts

​

Dgraph architecture

• Distributed by design: Automatically shards data across multiple nodes
• ACID transactions: Full transactional consistency across the distributed system
• GraphQL native: Direct GraphQL support without translation layers
• Type system: Strong typing with schema validation

​

Core concepts

• Predicates: Properties or relationships between nodes (similar to edges)
• UIDs: Unique identifiers for nodes (automatically managed)
• Facets: Properties on predicates (metadata about relationships)
• Types: Schema definitions that group predicates
• Indexes: Optimizations for specific query patterns

​

Graph data modeling

​

Step 1: Create your Hypermode Graph instance

​

Project structure

• Articles: News articles with content, metadata, and relationships
• Entities: People, organizations, locations mentioned in articles
• Topics: Subject categories and themes
• Sources: News outlets and publishers
• Temporal data: Publication dates and time-based relationships

​

Create your Hypermode graph

1

Sign in to Hypermode

Navigate to Hypermode and sign in.

2

Create a new graph

Select the “Develop” tab and click “Create Graph”. Give your graph a name and select “Create Graph”.

3

Copy the Dgraph connection string

Copy the dgraph:// connection string, you’ll use this to connect to your Hypermode Graph instance.

4

Open the Ratel interface

Navigate to ratel.hypermode.com and connect to your Hypermode Graph instance using the dgraph:// connection string

5

Update the graph schema

Select “Schema” from Ratel’s left navigation, then select “Bulk Edit” and copy and paste the following graph schema to replace the default schema, then select “Apply

Copy this DQL schema to update your graph schema

6

Import the sample data into your graph

Load RDF data into your graph, navigate to the Ratel interface and connect to your graph using the Dgraph connection string. Then run the following mutation in Ratel (be sure to select the Mutate tab):

Copy this DQL mutation to load sample data

​

Verify your graph

{
  articles(func:type(Article),first:100) {
    Article.title
    Article.uri
    Article.url
    Article.published
    Article.abstract
    Article.topic {
      Topic.name
    }
    Article.org {
      Organization.name
    }
    Article.geo {
      Geo.name
    }
  }
}

​

What you’ve accomplished

​

Next steps

​

Pro tip for today

Help me understand when to use different knowledge graph types:

1. For a customer support system: What type of knowledge graph would be most effective?
2. For a research assistant: How would you model academic papers and citations?
3. For a business intelligence system: What relationships matter most for competitive analysis?
4. For a content recommendation system: How do you balance multiple types of similarity?

Show me schema examples and explain the reasoning behind each design choice.

• Skills learned Dgraph architecture, RDF data modeling, multi-type knowledge graphs, temporal and provenance patterns, agent integration strategies