N-grams in NLP - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output of this code generating bigrams?

Given the code below that generates bigrams from a sentence, what is the output?

NLP

sentence = "I love machine learning"
words = sentence.split()
bigrams = [(words[i], words[i+1]) for i in range(len(words)-1)]
print(bigrams)

A [('I', 'love'), ('love', 'machine'), ('learning', 'machine')]

B [('I', 'love'), ('love', 'machine'), ('machine', 'learning')]

C [('love', 'I'), ('machine', 'love'), ('learning', 'machine')]

D [('I', 'machine'), ('love', 'learning')]

💡 Hint

Remember bigrams are pairs of consecutive words.

The code splits the sentence into words and pairs each word with the next one, producing consecutive word pairs (bigrams).

🧠 Conceptual

intermediate

Which statement best describes the purpose of n-grams in text processing?

Choose the best description of why n-grams are used in natural language processing.

A They capture sequences of words to understand context and word order.

B They remove stop words to reduce noise in text data.

C They translate text from one language to another.

D They count the total number of characters in a text.

💡 Hint

Think about how n-grams help capture word relationships.

N-grams group words into sequences to help models learn context and order, which is important for understanding meaning.

❓ Hyperparameter

advanced

Choosing the right n for n-grams

If you want to capture longer phrases but avoid very sparse data, which n-gram size is usually the best choice?

A Unigrams (n=1) because they are simple and cover all words.

B Four-grams (n=4) because they capture the most detailed phrases.

C Bigrams (n=2) because they balance context and data sparsity.

D Trigrams (n=3) because longer sequences always improve accuracy.

💡 Hint

Longer n-grams capture more context but can cause data sparsity.

Bigrams often provide a good balance between capturing context and keeping data manageable without too much sparsity.

❓ Metrics

advanced

Evaluating n-gram language models

Which metric is commonly used to evaluate the quality of an n-gram language model?

A Perplexity, which measures how well the model predicts a sample.

B Accuracy, which counts correct word predictions only.

C Mean Squared Error, used for regression tasks.

D F1 Score, used for classification balance.

💡 Hint

This metric measures uncertainty in predicting text sequences.

Perplexity measures how surprised a model is by the test data; lower perplexity means better prediction.

🔧 Debug

expert

Why does this n-gram code raise an error?

Consider this code snippet to generate trigrams. Why does it raise an IndexError?

NLP

sentence = "Data science is fun"
words = sentence.split()
trigrams = [(words[i], words[i+1], words[i+2]) for i in range(len(words)-2)]
print(trigrams)

A The split method does not create a list of words.

B The print statement is missing parentheses.

C Tuples cannot have three elements in Python.

D The range goes too far, causing words[i+2] to exceed list length.

💡 Hint

Check the range limit for accessing words[i+2].

The loop tries to access words[i+2] even when i is near the end, causing an IndexError.