To conduct sentiment analysis there are several different approaches that can be used to categorize the polarization of the tweets $($as either positive or negative$).$ To get the highest rate of accuracy, which type of model would probably be best?

Select one:

a$.$ One that uses an unsupervised learning approach.

b$.$ One that uses parts of speech tagging and sentence diagramming.

c$.$ One that uses the bag$-$of$-$words approach.

d$.$ One that uses only pre$-$defined rule$-$sets for classifying phrases as positive or negative.