Lowercasing and normalization in NLP - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

What is the output of this normalization code?

Consider the following Python code that normalizes a text by lowercasing and removing punctuation. What is the printed output?

NLP

import string
text = "Hello, World! Let's normalize this text."
normalized = ''.join(ch for ch in text.lower() if ch not in string.punctuation)
print(normalized)

A Hello World Lets Normalize This Text

B hello, world! let's normalize this text.

C hello world lets normalize this text

D hello world let's normalize this text

💡 Hint

Think about what lowercasing and removing punctuation does to the original text.

The code converts all letters to lowercase and removes all punctuation characters, including commas, apostrophes, and periods.

🧠 Conceptual

intermediate

Why is lowercasing important in text normalization?

Which of the following best explains why lowercasing is a common step in text normalization for machine learning?

A It increases the length of the text to improve model accuracy.

B It removes all punctuation from the text.

C It translates text into a different language.

D It reduces the number of unique words by treating 'Apple' and 'apple' as the same word.

💡 Hint

Think about how case differences affect word counts.

Lowercasing helps reduce vocabulary size by treating words that differ only in case as the same, which simplifies the model's learning.

❓ Metrics

advanced

How does normalization affect model accuracy?

You train two text classification models: Model A uses raw text, Model B uses normalized text (lowercased, punctuation removed). Which outcome is most likely?

A Model A achieves higher accuracy because raw text has more information.

B Model B achieves higher accuracy because normalization reduces noise and vocabulary size.

C Both models have the same accuracy because normalization does not affect text data.

D Model B performs worse because removing punctuation removes important meaning.

💡 Hint

Consider how noise and vocabulary size affect learning.

Normalization reduces noise and vocabulary size, helping the model generalize better and often improving accuracy.

🔧 Debug

advanced

Identify the error in this normalization code

What error does this code raise when run? import string text = "Normalize THIS!" normalized = ''.join(ch for ch in text.lower() if ch != string.punctuation) print(normalized)

A No error, prints 'normalize this!'

B TypeError: 'in <string>' requires string as left operand, not 'str'

C TypeError: '!=' not supported between instances of 'str' and 'str'

D The code prints 'normalize this' without punctuation

💡 Hint

Look carefully at how punctuation is checked in the condition.

No runtime error is raised. The condition 'ch != string.punctuation' always evaluates to True because a single character never equals the entire 'string.punctuation' string, so no punctuation is removed. Use 'ch not in string.punctuation' instead.

❓ Model Choice

expert

Choosing the best normalization for noisy text data

You have a dataset of social media posts with many uppercase letters, emojis, and punctuation. Which normalization approach is best before training a sentiment analysis model?

A Lowercase all text, remove punctuation, and remove emojis

B Keep original casing, keep punctuation, remove emojis

C Lowercase all text, keep punctuation and emojis

D Remove punctuation and emojis, keep original casing

💡 Hint

Consider what noise elements can confuse the model and what information is useful.

Lowercasing reduces vocabulary size, removing punctuation and emojis reduces noise that may not help sentiment detection, improving model focus.