Bag of Words

Also known as · Bag-of-Words · BoW

Describes each document as a vector of word counts and throws away word order, so "The food was great" and "Great was the food" give the same bag. Stacking these vectors gives the Document-Term Matrix.

When to use

It is the step after Text Preprocessing that turns cleaned tokens into numbers a model can use. Know its limitation: removing stopwords can destroy negation, so "not good" looks like "good", and frequency is not importance, which motivates TF-IDF.

See Lec 7.

Appears in

No references yet.