235,165 Turkish-language product reviews categorized into binary sentiment classes. The collection includes 220,284 positive and 14,881 negative labels mapped to the 'sentence' and 'sentiment' columns.
Use Cases
- Train a binary sentiment classifier using the 'sentence' text and 'sentiment' label
- Evaluate the performance of Turkish language models on imbalanced classification tasks due to the 14:1 positive-to-negative ratio
- Perform text mining or keyword extraction on the 'sentence' column to identify common product complaints or praises in Turkish
Strengths
- 235,165 total rows of Turkish product review data
- Binary sentiment labeling with 220,284 positive and 14,881 negative instances
- Simple schema consisting of 'sentence' (text) and 'sentiment' (integer) columns