IRLab_DAIICT at SemEval-2020 Task 9: Machine Learning and Deep Learning Methods for Sentiment Analysis of Code-Mixed Tweets

AbstractThe paper describes systems that our team IRLab_DAIICT employed for the shared task Sentiment Analysis for Code-Mixed Social Media Text in SemEval 2020. We conducted our experiments on a Hindi-English CodeMixed Tweet dataset which was annotated with sentiment labels. F1-score was the official evaluation metric and our best approach, an ensemble of Logistic Regression, Random Forest and BERT, achieved an F1-score of 0.693.