Vector Search

Movie Plots

Each Movie node in the database has a .plot property.

MATCH (m:Movie {title: "Toy Story"})
RETURN m.title AS title, m.plot AS plot

"A cowboy doll is profoundly threatened and jealous when a new spaceman figure supplants him as top toy in a boy's room."

Embeddings have been created for 1000 movie plots. The embedding is stored in the .plotEmbedding property of the Movie nodes.

MATCH (m:Movie {title: "Toy Story"})
RETURN m.title AS title, m.plot AS plot, m.plotEmbedding

A vector index, moviePlots, has been created for the .plotEmbedding property of the Movie nodes.

You can use the moviePlots vector index to find the most similar movies by comparing embeddings of movie plots.

Similarity Search

The Neo4jVector class provides an interface to use vector indexes in Neo4j. You can use Neo4jVector to create a vector store that can modify data and perform similarity search.

Open the genai-integration-langchain/vector_search.py file.

import os
from dotenv import load_dotenv
load_dotenv()

from langchain_neo4j import Neo4jGraph

# Connect to Neo4j
graph = Neo4jGraph(
    url=os.getenv("NEO4J_URI"),
    username=os.getenv("NEO4J_USERNAME"), 
    password=os.getenv("NEO4J_PASSWORD"),
)

# Create the embedding model

# Create Vector

# Search for similar movie plots

# Parse the documents

To perform a similarity search, you need to:

Embedding model

The movie plot embeddings were created using the OpenAI text-embedding-ada-002 model. You need to use the same model to convert the query into vectors.

Use the OpenAIEmbeddings class to create the embedding model:

from langchain_openai import OpenAIEmbeddings

# Create the embedding model
embedding_model = OpenAIEmbeddings(model="text-embedding-ada-002")

Vector Store

Use the Neo4jVector class to create a vector store that connects to the Neo4j database, uses the embedding model, and the moviePlots index.

from langchain_neo4j import Neo4jVector

# Create Vector
plot_vector = Neo4jVector.from_existing_index(
    embedding_model,
    graph=graph,
    index_name="moviePlots",
    embedding_node_property="plotEmbedding",
    text_node_property="plot",
)

When specifying the vector index you must also state the properties that contain the text (text_node_property) and the embedding (embedding_node_property).

Search

The similarity_search method of the Neo4jVector class allows you to perform a similarity search based on a query.

# Search for similar movie plots
plot = "Toys come alive"
result = plot_vector.similarity_search(plot, k=3)
print(result)

The query is converted into a vector using the embedding model, and then the vector index is used to find the most similar nodes.

The k parameter specifies the number of similar nodes to return.

import os
from dotenv import load_dotenv
load_dotenv()

from langchain_neo4j import Neo4jGraph
from langchain_neo4j import Neo4jVector
from langchain_openai import OpenAIEmbeddings

# Connect to Neo4j
graph = Neo4jGraph(
    url=os.getenv("NEO4J_URI"),
    username=os.getenv("NEO4J_USERNAME"), 
    password=os.getenv("NEO4J_PASSWORD"),
)

# Create the embedding model
embedding_model = OpenAIEmbeddings(model="text-embedding-ada-002")

# Create Vector
plot_vector = Neo4jVector.from_existing_index(
    embedding_model,
    graph=graph,
    index_name="moviePlots",
    embedding_node_property="plotEmbedding",
    text_node_property="plot",
)

# Search for similar movie plots
plot = "Toys come alive"
result = plot_vector.similarity_search(plot, k=3)
print(result)

Running the code will return the most similar movies to the query.

The method returns a list of LangChain Document objects, each containing the plot as the content and the node properties as metadata.

You can parse the results to extract the movie titles and plots.

# Parse the documents
for doc in result:
    print(f"Title: {doc.metadata['title']}")
    print(f"Plot: {doc.page_content}\n")

Experiment with different plots, such as:

result = plot_vector.similarity_search(
    plot,
    k=3,
    filter={"revenue": {"$gte": 200000000}}
)