GraphAcademy

Building Knowledge Graphs with LLMs

Knowledge graphs

LLM Graph Builder

Build a Graph with Python

Querying Knowledge Graphs

Explore a Knowledge Graph

The sandbox contains a prebuilt knowledge graph.

The knowledge graph represents just 3 news articles covering the 1976 United States presidential election:

  • 1976-6: "Jackson wins Massachusetts Democratic primary."

  • 1976-8: "Carter wins Pennsylvania Democratic primary"

  • 1976-22: "Jimmy Carter wins Presidency"

Learn more about these news articles

The 3 articles were taken from the NewsWire dataset that contains 2.7 million unique public domain U.S. news wire articles, written between 1878 and 1977.

The dataset was created to provide researchers with a large, high-quality corpus of historical news articles. These texts provide a massive repository of information about historical topics and events - and which newspapers were covering them. The dataset will be useful to a wide variety of researchers including historians, other social scientists, and NLP practitioners.

You can view the Python code which extracted these articles from the dataset in the llm-knowledge-graph-construction repository.

The knowledge graph maps the relationships between the following entity types referred to in the articles:

  • Person

  • Location

  • Organization

  • Building

  • Political party

  • State

Run this Cypher to reveal how the entities from the articles are related to each other:

cypher
MATCH p=(a:Article)-[:HAS_ENTITY]->(e)-[r]-()
RETURN p
A graph showing the entities the graph

You can see that all 3 articles are connected within the graph through the Person entity Jimmy Carter.

Node colors and captions

You can set the color of nodes and the text displayed by clicking on the node label, selecting a color and a property to use as the caption.

Screenshot of the node color and caption settings

The relationships within knowledge graphs allow you to explore how entities are connected and related to each other.

For example, how was Jimmy Carter connected to other Person entities:

cypher
MATCH (p:Person {id:"Jimmy Carter"})-[r]-(p2:Person)
RETURN p, r, p2
A graph showing how jimmy carter is connected to other people

You can use the nodes and relationships to understand how different entities are related, for example, Person and State entities:

cypher
MATCH (p:Person)-[r]-(s:State)
RETURN p, r, s

The following query will show you how the entities of a specific article are connected in the knowledge graph:

cypher
MATCH (a:Article {id:"1976-8"})-[:HAS_ENTITY]->(e)

MATCH (e)-[r]-(e2)
WHERE (a)-[:HAS_ENTITY]->(e2)

RETURN e, r, e2

Restricting the entities to only those from a specific article will give you a structured data view of that article.

Take some time to explore the knowledge graph and see how the entities are connected. Clicking on a node will display its properties. You can double-click on a node or click to focus on a node, then click the graph icon to expand its relationships.

Check Your Understanding

The Challenger to Jimmy Carter

Using the relationships in the knowledge graph determine who was the challenger to Jimmy Carter?

  • ❏ Sen Henry M. Jackson

  • ❏ Rep Morris K Udall

  • ✓ Gerald Ford

  • ❏ Walter F. Mondale

Hint

The answer is the id property on the (:Person) node at the end of the CHALLENGER relationship from Jimmy Carter.

Solution

The answer is Gerald Ford. You can find the answer by executing the following Cypher statement:

cypher
MATCH (p:Person)-[r]-(:Person {id:"Jimmy Carter"})
RETURN p.id AS Person, type(r) AS Relationship

Summary

In this lesson, you explored a pre-built knowledge graph in Neo4j.

In the next lesson, you will learn more about typical use cases for a knowledge graph.