Week 4: Text representation

DATA 441/641 · Week 4

Notes

• 4.1 Text Representation: Introduction(slides)
• 4.2 One-Hot Vectors(slides)
• 4.3 Bag of Words(slides)
• 4.4 Bag of N-Grams(slides)
• 4.5 TF-IDF(slides)

Labs

• Lab 4: Vectorizing Tweets(PDF)
• Lab 4 solutions (password)

Homework

• Homework 3(PDF)

Additional readings

• Practical NLP, Ch. 3 (text representation)
• sklearn: TfidfVectorizer

Discussion

• Why is a one-hot vector a poor measure of word similarity?
• When do bigrams help a classifier more than unigrams?