You are provided with a document collection where each document consists

of a text and a couple of keywords chosen from a predefined list. Each

document is categorized according to its topic $($e$.$g$.,$folk tales$$ or $$ancient

Greek comedy$).$ You trained a Naive Bayes classifier and a logistic

regression model using both keywords and text and observed that both

classifiers achieve very high accuracy. You also noticed that the keywords

on their own are highly predictive of the topic. Now, you are about to

demonstrate your system and learn that only text will be provided as input

but no keywords. You do not have time to train a new classifier; which of

the classifiers you previously trained should you use and why?