Stopwords
Very common, low-meaning words (the, is, at, was, a, and) removed before analysis to focus on the content words; it typically cuts the token count by 40 to 60 percent.
When to use
A step in the preprocessing pipeline after Tokenization. Watch the pitfall the lecture flags: not is often a stopword, so removal can destroy negation ("not good" loses its meaning), a known limit of Bag of Words. Different libraries ship different lists. Also in PS 06.
See DS Lec 7.