Datasets:
| language: | |
| - en | |
| language_creators: | |
| - found | |
| license: unknown | |
| multilinguality: | |
| - monolingual | |
| pretty_name: Blue | |
| size_categories: | |
| - 10K<n<100K | |
| source_datasets: | |
| - extended|app_reviews | |
| tags: | |
| - reviews | |
| - ratings | |
| - ordinal | |
| - text | |
| task_categories: | |
| - text-classification | |
| task_ids: | |
| - text-scoring | |
| - sentiment-scoring | |
| This dataset is a quick-and-dirty benchmark for predicting ratings across | |
| different domains and on different rating scales based on text. It pulls in a | |
| bunch of rating datasets, takes at most 1000 instances from each and combines | |
| them into a big dataset. | |
| Requires the `kaggle` library to be installed, and kaggle API keys passed | |
| through environment variables or in ~/.kaggle/kaggle.json. See [the Kaggle | |
| docs](https://www.kaggle.com/docs/api#authentication). | |