README.md

---
annotations_creators:
- Duygu Altinok
language:
- tr
license:
- cc-by-sa-4.0
multilinguality:
- monolingual
size_categories:
- 100K<n<1M
source_datasets:
- original
task_categories:
- text-classification
task_ids:
- sentiment-classification
pretty_name: SentiTurca (Sentiment Analysis Datasets for Turkish language)
config_names:
- e-commerce
- hate
- movies
tags:
- sentiment
dataset_info:
- config_name: hate
  features:
  - name: baslik
    dtype: string
  - name: text
    dtype: string
  - name: label
    dtype:
      class_label:
        names:
          0: offensive
          1: hate
          2: neutral
          3: civilized
  splits:
  - name: train
    num_bytes: 47357639
    num_examples: 42175
  - name: validation
    num_bytes: 5400927
    num_examples: 5000
  - name: test
    num_bytes: 5323545
    num_examples: 5000
  download_size: 58918801
- config_name: movies
  features:
  - name: text
    dtype: string
  - name: label
    dtype: 
      class_label:
        names:
          0: negative
          1: positive
  splits:
  - name: train
    num_bytes: 46979645
    num_examples: 60411
  - name: validation
    num_bytes: 733500
    num_examples: 8905
  - name: test
    num_bytes: 742661
    num_examples: 8934
  download_size: 58918801
- config_name: e-commerce
  features:
  - name: text
    dtype: string
  - name: label
    dtype:
      class_label:
        names:
          0: 1_star
          1: 2_star
          2: 3_star
          3: 4_star
          4: 5_star
  splits:
  - name: train
    num_bytes: 12844466
    num_examples: 73920
  - name: validation
    num_bytes: 4811620
    num_examples: 15000
  - name: test
    num_bytes: 5260694
    num_examples: 15000
configs:
- config_name: movies
  data_files:
  - split: train
    path: movies/train-*
  - split: validation
    path: movies/validation-*
  - split: test
    path: movies/test-*
- config_name: e-commerce
  data_files:
  - split: train
    path: e-commerce/train*
  - split: validation
    path: e-commerce/valid*
  - split: test
    path: e-commerce/test*
- config_name: hate
  data_files:
  - split: train
    path: hate/train-*
  - split: validation
    path: hate/validation-*
  - split: test
    path: hate/test-*
---

# SentiTurca - A Sentiment Analysis Benchmark for Turkish


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



# Dataset Card for SentiTurca

SentiTurca is a sentiment analysis benchmarking dataset including movie reviews, hate speech and e-commerce reviews classification.  

### Datasets
**e-commerce**: The e-commerce reviews are scraped from e-commerce websites Trendyol.com and Hepsiburada.com, including review for many product types such as cloths, toys, books, electronics and more.  
E-commerce reviews has their [stand alone HF repo](https://huggingface.co/datasets/turkish-nlp-suite/MusteriYorumlari) as well.

**movies** The movie reviews are scraped from two movie review websites, Sinefil.com and Beyazperde.com. Here, we used 2 labels but for a total challenge of 10 label classification can be found under this dataset's [stand alone HF repo](https://huggingface.co/datasets/turkish-nlp-suite/BuyukSinema).
This dataset is also a part of [TrGLUE benchmark](https://huggingface.co/datasets/turkish-nlp-suite/TrGLUE) under the task name **sst2**.

**hate** This dataset is the [Turkish Hate Map](https://huggingface.co/datasets/turkish-nlp-suite/TurkishHateMap), scraped from Eksisozluk.com and including 4 labels: offense, hate, neutral and civilized.


### Dataset statistics

Here are the dataset sizes and number of labels:


| Subset | size | num labels |
|---|---|---|
| e-commerce  | 103K | 5 |
| movies  | 78K  | 2|
| hate |  52K| 4 |


### Benchmarking

We benchmarked BERTurk on all of our datasets. 
All benchmarking scripts can be found under the dedicated [SentiTurca Github repo](https://github.com/turkish-nlp-suite/SentiTurca).

| Subset | metrics | success |
|---|---|---|
| movies  | Matthews corr. | 0.67 |
| e-commerce  | acc./F1 | 0.66/0.64 |
| hate | acc./F1 | 0.61/0.58 | 

As one sees, hate dataset is quite challenging. For a  full critique of the benchmark please visit our [research paper]().


### Citation

Coming soon!