---
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
dataset_info:
  features:
  - name: review
    dtype: string
  - name: label
    dtype: string
  - name: topic
    dtype: string
  - name: writing_style
    dtype: string
  splits:
  - name: test
    num_bytes: 109030
    num_examples: 851
  download_size: 50837
  dataset_size: 109030
task_categories:
- text-classification
size_categories:
- n<1K
---
# darija-reviews


**Description**: This small dataset consists of product and service reviews from social media written in Darija, encompassing both Arabic and Arabizi writing styles. 
Each review is categorized by its polarity (positive, negative, or neutral) and includes information about the domain or topic, as well as the writing style of the content. 
The dataset encompasses a broad spectrum of topics, including clothing, cosmetics, entertainment, hospitality, IT, and other domains.

**Size**: The dataset contains a total of 851 reviews.

**Labels**:

- Sentiment Labels: Positive, Negative, Neutral  
- Topics: Automotive, Cleaning, Clothing, Cosmetics, Entertainment, Hospitality, Household Appliances, IT, Jewelry, Restaurants, and Other.  
- Writing Styles: Arabic & Arabizi  

**Usage**: This dataset can be used for ***evaluating*** the performance of sentiment analysis models in classifying the polarity (positive, negative, or neutral) of product and service reviews written in Darija.

       
**Examples**:

- Positive IT Review (Arabic WS): "هاتف زوين عجبني"  
- Negative Cosmetics Review (Arabizi WS): "Lwa7iid li fach kandirou wejhi kayt9lab blhboub manhar 9te3to ou lhbob hbso hada maysla7ch lbchra dohniya myidem bzzaf"  
- Neutral Cleaning Review (Arabic WS): "خليه شوية على الطبيع ومسحي بشيفون فازك"