Stemming (Porter, Snowball) in NLP - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of Porter Stemmer on a word list

What is the output list after applying the Porter Stemmer to the words ['running', 'jumps', 'easily', 'fairly']?

NLP

from nltk.stem import PorterStemmer
ps = PorterStemmer()
words = ['running', 'jumps', 'easily', 'fairly']
stemmed = [ps.stem(word) for word in words]
print(stemmed)

A ['run', 'jump', 'easily', 'fairly']

B ['running', 'jump', 'easily', 'fairly']

C ['run', 'jumps', 'easili', 'fairly']

D ['run', 'jump', 'easili', 'fairli']

💡 Hint

Porter Stemmer often removes suffixes like 'ing', 's', and changes 'y' endings.

Porter Stemmer reduces words to their root form by removing common suffixes. 'running' becomes 'run', 'jumps' becomes 'jump', 'easily' becomes 'easili', and 'fairly' becomes 'fairli'.

🧠 Conceptual

intermediate

Difference between Porter and Snowball Stemmer

Which statement correctly describes a key difference between the Porter Stemmer and the Snowball Stemmer?

A Snowball Stemmer always produces longer stems than Porter Stemmer.

B Porter Stemmer supports multiple languages while Snowball only supports English.

C Snowball Stemmer is a newer, more readable and flexible version of Porter Stemmer supporting multiple languages.

D Porter Stemmer uses machine learning while Snowball uses rule-based stemming.

💡 Hint

Think about language support and code design improvements.

Snowball Stemmer was designed as an improved version of Porter Stemmer with clearer rules and support for many languages, unlike Porter which is older and English-only.

❓ Metrics

advanced

Evaluating stemming impact on text classification accuracy

You train a text classifier on raw text and then on stemmed text using Porter Stemmer. The accuracy on test data changes from 82% to 79%. What is the most likely explanation?

A Stemming always improves accuracy, so this must be a bug in the code.

B Stemming reduced vocabulary size but also removed useful distinctions, slightly lowering accuracy.

C The test data was stemmed but training data was not, causing mismatch and accuracy drop.

D Porter Stemmer introduced spelling errors that confused the classifier.

💡 Hint

Consider how stemming affects word forms and model learning.

Stemming reduces word forms to roots, which can reduce vocabulary and noise but may also remove meaningful differences, sometimes slightly lowering accuracy.

🔧 Debug

advanced

Identifying error in Snowball Stemmer usage

What error will this code raise? from nltk.stem import SnowballStemmer stemmer = SnowballStemmer('english') print(stemmer.stem(123))

A TypeError because stem() expects a string, not an integer

B AttributeError because integers have no lower() method

C ValueError because 'english' is not a valid language

D No error, outputs '123'

💡 Hint

Check the input type expected by stem() method.

SnowballStemmer.stem() expects a string input. Passing an integer causes a TypeError.

❓ Model Choice

expert

Choosing stemming method for multilingual text preprocessing

You have a dataset with English, Spanish, and French texts. Which stemming approach is best to preprocess this data before training a model?

A Use Snowball Stemmer specifying the language for each text before stemming

B Use Porter Stemmer on all texts regardless of language

C Use a custom rule-based stemmer designed only for English

D Use no stemming and rely on raw text for all languages

💡 Hint

Consider language support in stemming tools.

Snowball Stemmer supports multiple languages and can be applied per language, making it suitable for multilingual datasets. Porter Stemmer only supports English.