Persian Customer Review Purchase Suggestion Classification
A Persian NLP project that classifies customer reviews into three categories:
- Buy: The customer recommends purchasing the product.
- Not Buy: The customer does not recommend purchasing the product.
- Neutral: The review does not express a clear purchase recommendation.
The project uses XLM-RoBERTa fine-tuning with PyTorch and Hugging Face Transformers to understand Persian customer feedback and predict purchase intent.
Project Goal
The goal is to automatically analyze Persian customer reviews and determine whether a customer would recommend buying a product.
This can help businesses:
- Monitor customer satisfaction
- Analyze product feedback
- Identify negative experiences
- Understand purchase intention from reviews
Classification Labels
The dataset contains three labels:
| Label | Meaning |
|---|
| 0 | Buy |
| 1 | Neutral |
| 2 | Not Buy |
Model
- Base model:
xlm-roberta-base
- Language: Persian / multilingual
- Framework: PyTorch
- Task: 3-class text classification
