Noisy data is data with a large amount of additional meaningless information in it (noise). It can be the most ______ tokens in the dataset.

I. frequent
II. sparse

Correct Answer: I and II

Data = true signal + noise. Noise is unwanted data items, features or records which don’t help in explaining the feature itself. Stop words can appear frequently but are irrelevant. Rare tokens are usually noise features that should be removed.