Information retrieval basics in NLP - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main purpose of the TF-IDF score in information retrieval?

TF-IDF is a common technique used in information retrieval. What does it primarily help with?

A It measures how important a word is to a document relative to a collection of documents.

B It counts the total number of words in a document.

C It ranks documents based on their length.

D It removes stop words from the text.

💡 Hint

Think about how TF-IDF balances word frequency with rarity across documents.

TF-IDF stands for Term Frequency-Inverse Document Frequency. It scores words higher if they appear often in a document but rarely across all documents, helping to find important words.

❓ Predict Output

intermediate

What is the output of this code computing cosine similarity?

Given two vectors representing documents, what is the cosine similarity output?

NLP

import numpy as np

def cosine_similarity(vec1, vec2):
    dot_product = np.dot(vec1, vec2)
    norm1 = np.linalg.norm(vec1)
    norm2 = np.linalg.norm(vec2)
    return dot_product / (norm1 * norm2)

vec1 = np.array([1, 2, 3])
vec2 = np.array([4, 5, 6])
result = cosine_similarity(vec1, vec2)
print(round(result, 2))

A 0.50

B 0.97

C 1.00

D 0.74

💡 Hint

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

Cosine similarity is the dot product of vectors divided by the product of their magnitudes. For these vectors, the value is approximately 0.9746, rounded to 0.97.

❓ Model Choice

advanced

Which model is best suited for semantic search in information retrieval?

You want to build a search system that understands the meaning of queries and documents beyond exact word matches. Which model would you choose?

A TF-IDF vectorizer

B Simple keyword matching

C Bag-of-Words model

D Pretrained transformer-based embeddings (e.g., BERT)

💡 Hint

Consider models that capture context and meaning, not just word counts.

Transformer-based models like BERT produce embeddings that capture semantic meaning, enabling better understanding of queries and documents for semantic search.

❓ Hyperparameter

advanced

Which hyperparameter affects the number of neighbors considered in k-NN for document retrieval?

In a k-Nearest Neighbors (k-NN) model used for retrieving similar documents, which hyperparameter controls how many neighbors are checked?

A batch size

B learning rate

C k (number of neighbors)

D max depth

💡 Hint

Think about the 'k' in k-NN and what it stands for.

The hyperparameter 'k' specifies how many nearest neighbors the model considers when making a retrieval or classification decision.

❓ Metrics

expert

Which metric best evaluates ranking quality in information retrieval?

You want to measure how well your search engine ranks relevant documents higher than irrelevant ones. Which metric is most appropriate?

A Normalized Discounted Cumulative Gain (NDCG)

B Mean Squared Error (MSE)

C Recall

D Precision at k (P@k)

💡 Hint

Consider metrics that account for both relevance and position in the ranked list.

NDCG measures ranking quality by considering the relevance of documents and their positions, rewarding relevant documents appearing higher in the list.