Table 2 Overview of datasets used for sentiment analysis experiments.
Dataset | Total reviews | App source(s) | Composition | Purpose |
|---|---|---|---|---|
Dataset 1 | 10,000 | All reviews from Instagram | Initial fine-tuning of smaller BERT | |
Dataset 2 | 20,000 | Instagram & Threads | 10,000 from Instagram, 10,000 from Threads | Main experiments (Standard BERT-Base & RoBERTa) |
Dataset 3 | 40,000 | Instagram & Threads | 20,000 from Instagram, 20,000 from Threads | Evaluation of large-scale performance |
