The "norm" parameter of TfidfVectorizer in scikit-learn is used to specify the normalization method applied to the resulting TF-IDF vectors. It controls whether the vectors should be normalized to have unit norm (L2 normalization) or left unnormalized (None).