---
license: apache-2.0
task_categories:
- text-classification
language:
- en
---
# TripAdvisor Review Rating Split Dataset

This dataset contains **80,000 TripAdvisor reviews** with corresponding **ratings**. It is derived from the original TripAdvisor dataset available [here](https://huggingface.co/datasets/jniimi/tripadvisor-review-rating) and was created to train different models for a university project in the class of NLP.

## Dataset Structure

- **Training Set:** 30,400 examples
- **Validation Set:** 1,600 examples
- **Test Set:** 8,000 examples

Each set is **balanced**, ensuring equal representation of all sentiment labels. 

## Label

The labels represent the **overall rating** from the original dataset. However, note that this label may not always fully align with the specific content or tone of the text within each review. The label is numeric, ranging from 1 to 5.

## Purpose

This dataset is intended to be used for training various models to predict the sentiment of TripAdvisor reviews based on their content. It is ideal for use in natural language processing tasks such as sentiment analysis and review classification.

## Limitation

The label used in this dataset, **overall rating**, may not always accurately reflect the detailed sentiment or tone expressed in the review text. Since the **overall rating** is a general assessment by the reviewer, it may not always align with the nuances or specificities of the content in the review.

## Copyright

This dataset is derived from the original TripAdvisor dataset provided by [jniimi](https://huggingface.co/jniimi). All rights to the original dataset are owned by the author. This dataset is intended for academic and non-commercial use only.