Similarity search and retrieval in Prompt Engineering / GenAI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main purpose of similarity search in AI?

Imagine you have a large photo album and want to find pictures that look like a specific photo. What does similarity search do in this context?

A It finds photos that are exactly the same as the given photo.

B It finds photos that share similar features or patterns with the given photo.

C It sorts all photos by their file size.

D It deletes photos that are not similar to the given photo.

💡 Hint

Think about how you find things that look alike, not exactly the same.

Similarity search helps find items that share common features or patterns, not just exact copies.

❓ Predict Output

intermediate

Output of cosine similarity calculation

What is the output of this cosine similarity calculation between vectors A and B?

Prompt Engineering / GenAI

import numpy as np
A = np.array([1, 2, 3])
B = np.array([4, 5, 6])
cos_sim = np.dot(A, B) / (np.linalg.norm(A) * np.linalg.norm(B))
print(round(cos_sim, 2))

A 0.97

B 1.00

C 0.87

D 0.75

💡 Hint

Recall cosine similarity formula and calculate dot product and norms carefully.

Cosine similarity measures the angle between two vectors; here it is approximately 0.87.

❓ Model Choice

advanced

Best model type for similarity search on text data

You want to build a system that finds similar sentences in a large document collection. Which model type is best suited for this task?

A K-means clustering on raw text tokens

B Recurrent Neural Network (RNN) trained for language modeling

C Convolutional Neural Network (CNN) trained for image classification

D Transformer-based model producing sentence embeddings

💡 Hint

Think about models that create meaningful vector representations of sentences.

Transformer models like BERT create embeddings that capture sentence meaning, ideal for similarity search.

❓ Metrics

advanced

Choosing the right metric for nearest neighbor search

Which distance metric is most appropriate for measuring similarity between high-dimensional dense vectors from a neural network embedding?

A Cosine similarity

B Manhattan distance

C Euclidean distance

D Hamming distance

💡 Hint

Consider which metric focuses on the angle between vectors rather than magnitude.

Cosine similarity measures the angle between vectors, which is effective for high-dimensional embeddings.

🔧 Debug

expert

Debugging a similarity search code snippet

What error does this code raise when trying to compute similarity between two lists of different lengths?

Prompt Engineering / GenAI

def jaccard_similarity(list1, list2):
    intersection = len(set(list1) & set(list2))
    union = len(set(list1) | set(list2))
    return intersection / union

result = jaccard_similarity([1, 2, 3], [1, 2])
print(result)

A No error, output is 0.67

B TypeError

C ZeroDivisionError

D IndexError

💡 Hint

Check how sets handle different length lists and the division operation.

The code correctly computes Jaccard similarity without error even if lists differ in length.