If you prefer the Movies dataset over Northwind, complete this challenge to create embeddings and build an agent for movies.
What You Will Do
-
Use your existing Movies dataset (Movie, Person, ACTED_IN) from Aura Fundamentals
-
Create embeddings on Movie nodes using
text-embedding-ada-002 -
Create a vector index for semantic search
-
Build an agent with tools for Movie and Person nodes
Create Embeddings for Movie Nodes
Run this Cypher in the Query tool. You need an OpenAI API key. The GenAI plugin is enabled by default in Aura.
MATCH (m:Movie)
WHERE m.title IS NOT NULL AND m.embedding IS NULL
CALL {
WITH m
WITH m, substring(coalesce(m.tagline, '') + "\n" + m.title, 0, 1000) AS movieText
WITH m, genai.vector.encode(
movieText,
"OpenAI",
{ token: "sk-...", model: "text-embedding-ada-002" }) AS vector
SET m.embedding = vector
} IN TRANSACTIONS OF 5 ROWS
ON ERROR CONTINUE
RETURN count(m) AS updatedReplace token: "sk-…" with your OpenAI API key.
Create the Vector Index
Create a vector index on Movie.embedding:
-
Dimensions: 1536 (for
text-embedding-ada-002) -
Similarity: cosine
Build Your Agent
Adapt the Northwind Analyst pattern for Movies:
-
Tools: Get Movie, Get Person, Movies by Actor, Actors in Movie, Similar Movies (vector search)
-
Instructions: Describe the Movie/Person schema and when to use each tool
-
Test questions: "Find movies similar to The Matrix", "Which actors appeared in action movies?"
Summary
You built an agent using the Movies dataset with ada embeddings. If you complete this path, you can skip the Northwind setup and proceed to the publishing challenge.