Semantic search for products

Time to put your find_n_closest() function to use! You'll test out your semantic product search on a test query, computing a sorted list of the five most semantically similar products, based on the enriched data you gave the model.

Here's a reminder of the find_n_closest() function you created in the previous exercise:

def find_n_closest(query_vector, embeddings, n=3):
    distances = []
    for index, embedding in enumerate(embeddings):
        distance = spatial.distance.cosine(query_vector, embedding)
        distances.append({"distance": distance, "index": index})
    distances_sorted = sorted(distances, key=lambda x: x["distance"])
    return distances_sorted[0:n]

The create_embeddings() function you created earlier is also available. Recall that it takes some text, and returns a list of lists containing the embeddings for each text. The products dictionary and the product_embeddings you created previously have also been loaded.

This exercise is part of the course

Introduction to Embeddings with the OpenAI API

View Course

Exercise instructions

Create the query vector from query_text.
Find the five closest distances and their corresponding indexes using the find_n_closest() function.
Loop over hits and extract the product at each 'index' in the hits list.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Create the query vector from query_text
query_text = "computer"
query_vector = ____[0]

# Find the five closest distances
hits = ____

print(f'Search results for "{query_text}"')
for hit in hits:
  # Extract the product at each index in hits
  product = products[____]
  print(product["title"])

Edit and Run Code