SpeechTrans@SMM4H’20: Impact of Preprocessing and N-grams on Automatic Classification of Tweets That Mention Medications

AbstractThis paper describes our system developed for automatically classifying tweets that mention medications. We used the Decision Tree classifier for this task. We have shown that using some elementary preprocessing steps and TF-IDF n-grams led to acceptable classifier performance. Indeed, the F1-score recorded was 74.58% in the development phase and 63.70% in the test phase.