Transfer Learning for Low-Resource Sentiment Analysis
This work addresses sentiment analysis for a low-resource language, but it is incremental as it applies existing methods to new data.
The paper tackles sentiment analysis for Central Kurdish by collecting a new dataset and applying transfer learning for data augmentation, achieving high F1 score and accuracy despite the task's difficulty.
Sentiment analysis is the process of identifying and extracting subjective information from text. Despite the advances to employ cross-lingual approaches in an automatic way, the implementation and evaluation of sentiment analysis systems require language-specific data to consider various sociocultural and linguistic peculiarities. In this paper, the collection and annotation of a dataset are described for sentiment analysis of Central Kurdish. We explore a few classical machine learning and neural network-based techniques for this task. Additionally, we employ an approach in transfer learning to leverage pretrained models for data augmentation. We demonstrate that data augmentation achieves a high F$_1$ score and accuracy despite the difficulty of the task.