Semantic text search using embeddings

Semantic text search using embeddings¶

We can search through all our reviews semantically in a very efficient manner and at very low cost, by simply embedding our search query, and then finding the most similar reviews. The dataset is created in the Obtain_dataset Notebook.

In [1]:

import pandas as pd
import numpy as np

datafile_path = "data/fine_food_reviews_with_embeddings_1k.csv"

df = pd.read_csv(datafile_path)
df["embedding"] = df.embedding.apply(eval).apply(np.array)

import pandas as pd import numpy as np datafile_path = "data/fine_food_reviews_with_embeddings_1k.csv" df = pd.read_csv(datafile_path) df["embedding"] = df.embedding.apply(eval).apply(np.array)

Remember to use the documents embedding engine for documents (in this case reviews), and query embedding engine for queries. Note that here we just compare the cosine similarity of the embeddings of the query and the documents, and show top_n best matches.

In [2]:

from openai.embeddings_utils import get_embedding, cosine_similarity

# search through the reviews for a specific product
def search_reviews(df, product_description, n=3, pprint=True):
    product_embedding = get_embedding(
        product_description,
        engine="text-embedding-ada-002"
    )
    df["similarity"] = df.embedding.apply(lambda x: cosine_similarity(x, product_embedding))

    results = (
        df.sort_values("similarity", ascending=False)
        .head(n)
        .combined.str.replace("Title: ", "")
        .str.replace("; Content:", ": ")
    )
    if pprint:
        for r in results:
            print(r[:200])
            print()
    return results


results = search_reviews(df, "delicious beans", n=3)

from openai.embeddings_utils import get_embedding, cosine_similarity # search through the reviews for a specific product def search_reviews(df, product_description, n=3, pprint=True): product_embedding = get_embedding( product_description, engine="text-embedding-ada-002" ) df["similarity"] = df.embedding.apply(lambda x: cosine_similarity(x, product_embedding)) results = ( df.sort_values("similarity", ascending=False) .head(n) .combined.str.replace("Title: ", "") .str.replace("; Content:", ": ") ) if pprint: for r in results: print(r[:200]) print() return results results = search_reviews(df, "delicious beans", n=3)

Good Buy:  I liked the beans. They were vacuum sealed, plump and moist. Would recommend them for any use. I personally split and stuck them in some vodka to make vanilla extract. Yum!

Jamaican Blue beans:  Excellent coffee bean for roasting. Our family just purchased another 5 pounds for more roasting. Plenty of flavor and mild on acidity when roasted to a dark brown bean and befor

Delicious!:  I enjoy this white beans seasoning, it gives a rich flavor to the beans I just love it, my mother in law didn't know about this Zatarain's brand and now she is traying different seasoning

In [3]:

results = search_reviews(df, "whole wheat pasta", n=3)

results = search_reviews(df, "whole wheat pasta", n=3)

Tasty and Quick Pasta:  Barilla Whole Grain Fusilli with Vegetable Marinara is tasty and has an excellent chunky vegetable marinara.  I just wish there was more of it.  If you aren't starving or on a 

sooo good:  tastes so good. Worth the money. My boyfriend hates wheat pasta and LOVES this. cooks fast tastes great.I love this brand and started buying more of their pastas. Bulk is best.

Handy:  Love the idea of ready in a minute pasta and for that alone this product gets praise.  The pasta is whole grain so that's a big plus and it actually comes out al dente.  The vegetable marinara

We can search through these reviews easily. To speed up computation, we can use a special algorithm, aimed at faster search through embeddings.

In [4]:

results = search_reviews(df, "bad delivery", n=1)

results = search_reviews(df, "bad delivery", n=1)

great product, poor delivery:  The coffee is excellent and I am a repeat buyer.  Problem this time was with the UPS delivery.  They left the box in front of my garage door in the middle of the drivewa

As we can see, this can immediately deliver a lot of value. In this example we show being able to quickly find the examples of delivery failures.

In [5]:

results = search_reviews(df, "spoilt", n=1)

results = search_reviews(df, "spoilt", n=1)

Extremely dissapointed:  Hi,<br />I am very disappointed with the past shipment I received of the ONE coconut water. 3 of the boxes were leaking and the coconut water was spoiled.<br /><br />Thanks.<b

In [6]:

results = search_reviews(df, "pet food", n=2)

results = search_reviews(df, "pet food", n=2)

Good food:  The only dry food my queen cat will eat. Helps prevent hair balls. Good packaging. Arrives promptly. Recommended by a friend who sells pet food.

The cats like it:  My 7 cats like this food but it is a little yucky for the human. Pieces of mackerel swimming in a dark broth. It is billed as a "complete" food and contains carrots, peas and pasta.