A new arXiv paper explores methods to improve sentiment classification performance on small datasets. The study compares three data augmentation techniques: Easy Data Augmentation (EDA), back-translation, and contextual token substitution using nlpaug. These methods were tested with both traditional classifiers like Logistic Regression and Random Forest, and transformer models such as BERT. The research found that all augmentation strategies enhanced model performance, with contextual augmentation showing the most consistent benefits for BERT, while EDA and back-translation were more effective for traditional models. AI
IMPACT Provides insights into selecting effective data augmentation techniques for sentiment classification tasks with limited data.
RANK_REASON The cluster contains an academic paper detailing research findings on data augmentation techniques for sentiment classification. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- Back-Translation for Cross-Cultural Research
- Bert
- Easy Data Augmentation
- Hugging Face
- logistic regression model
- Mahmoud S. Mayaleh
- nlpaug
- random forest
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
