We present a transfer learning system to perform a mixed Spanish-Englishsentiment classification task . Our proposal uses the state-of-the-art languagemodel BERT and embeds it within a ULMFiT transfer learning pipeline . Thiscombination allows us to predict the polarity detection of code-mixed tweets . Among 29 submitted systems, our approach is ranked 4th on the Sentimix Spanglish test set ofSemEval 2020 Task 9 .

Author(s) : Daniel Palomino, Jose Ochoa-Luna

Links : PDF - Abstract

Code :

Keywords : task - mixed - code - transfer - learning -

Leave a Reply

Your email address will not be published. Required fields are marked *