Hello, i have a question about the impact of ponctuation on the prediction.
I have this sentence in my training data: “comment me rendre à illkirch?”
I have used stop_words for CountVectorsFeaturizer to remove some stop_words. As i have seen in the pkl file, the ponctuation is not taken into consideration.
In my test, i have tested with and without ponctuation and the results are not the same
example 1 : “me rendre illkirch” 0.81 Acces_plan example 2 : “me rendre illkirch?” 0.94 Acces_plan
Any explication please ?