turkish-nlp-suite/BuyukSinema

---
annotations_creators:
- Duygu Altinok
language:
- tr
license:
- cc-by-sa-4.0
multilinguality:
- monolingual
size_categories:
- 10K<n<100K
source_datasets:
- original
task_categories:
- text-classification
task_ids:
- sentiment-classification
pretty_name: BuyukSinema
tags:
- sentiment
dataset_info:
features:
- name: text
dtype: string
- name: label
dtype:
class_label:
names:
'0': nono
'1': nowatch
'2': horrible
'3': poor
'4': bad
'5': middle
'6': good
'7': great
'8': super
'9': amazing
splits:
- name: train
num_bytes: 46979645
num_examples: 67328
- name: validation
num_bytes: 733500
num_examples: 10000
- name: test
num_bytes: 742661
num_examples: 10000
download_size: 58918801
data_files:
- split: train
path: movies/train-*
- split: validation
path: movies/validation-*
- split: test
path: movies/test-*
---

# BüyükSinema - A Large Scale Turkish Movie Reviews Sentiment Dataset

<img src="https://raw.githubusercontent.com/turkish-nlp-suite/.github/main/profile/buyuksinema.png" width="30%" height="30%">

## Dataset Summary

BüyükSinema is a Turkish movie reviews dataset of size 87K, scraped from Sinefil.com and Beyazperde.com. Hence this dataset is a superset of
[BeyazPerde All Movie Reviews](https://huggingface.co/datasets/turkish-nlp-suite/beyazperde-all-movie-reviews),
[BeyazPerde Top 300 Movie Reviews](https://huggingface.co/datasets/turkish-nlp-suite/beyazperde-top-300-movie-reviews) and
[Sinefil Movie Reviews](https://huggingface.co/datasets/turkish-nlp-suite/sinefil-movie-reviews) datasets.

This is a merge of the three different datasets from two resources, hence we scaled the output stars into the range of 1-10 accordingly.

The star distribution is as follows:

| star rating | count |
|---|---|
| 1 | 5,657 |
| 2 | 3,092 |
| 3 | 2,172 |
| 4 | 3,491 |
| 5 | 7,349 |
| 6 | 9,078 |
| 7 | 15,647 |
| 8 | 21,154 |
| 9 | 10,868 |
| 10 | 8,820 |
| total | 87,328 |

The star distribution is quite skewed towards 7+ stars. For more information about dataset statistics, please refer to the [research paper]().

## Dataset Instances

An instance looks like:

```
{
"text":"Mükemmelin ötesinde bir şey. Helal olsun. Devamını da isteriz artık... Emeğinize Yüreğinize Sağlık...",
"label":9
}
```

## Data Split

| name |train|validation|test|
|---------|----:|---:|---:|
|BüyükSinema Movie Reviews|67328|10000|10000|

## Benchmarking
This dataset is a part of [TRGLUE](https://huggingface.co/datasets/turkish-nlp-suite/TrGLUE) and [SentiTurca](https://huggingface.co/datasets/turkish-nlp-suite/SentiTurca) benchmarks, in the benchmark the subset name is **TrSST-2**, named according to the GLUE tasks.
Also the TrGLUE and SentiTurca tasks are binary classification tasks to follow original GLUE conventions.
In this repo, you can access the original star ratings if you want a challenge.

We benchmarked the transformer based model BERTurk on the binary classification task, this model achieved a **0.67** Matthews's correlation coefficient.
More information can be found in the [research paper]() and benchmarking code can be found under [TrGLUE Github repo](https://github.com/turkish-nlp-suite/TrGLUE).

## Citation
Coming soon!!