Performance of Data Augmentation Methods for Brazilian Portuguese Text Classification
This work addresses the lack of research on data augmentation for non-English languages like Brazilian Portuguese, though it is incremental as it applies known methods to new data.
The study evaluated existing data augmentation methods for text classification on Brazilian Portuguese corpora, finding some improvements but highlighting issues with language bias and data scarcity.
Improving machine learning performance while increasing model generalization has been a constantly pursued goal by AI researchers. Data augmentation techniques are often used towards achieving this target, and most of its evaluation is made using English corpora. In this work, we took advantage of different existing data augmentation methods to analyze their performances applied to text classification problems using Brazilian Portuguese corpora. As a result, our analysis shows some putative improvements in using some of these techniques; however, it also suggests further exploitation of language bias and non-English text data scarcity.