https://github.com/praths71018/hindi_sentiment_analysis

Understanding the sentiment of customers from product reviews using IndicBERT
https://github.com/praths71018/hindi_sentiment_analysis

hindi indicbert machine-learning nlp product-reviews sentiment-analysis

Last synced: 7 months ago
JSON representation

Understanding the sentiment of customers from product reviews using IndicBERT

Host: GitHub
URL: https://github.com/praths71018/hindi_sentiment_analysis
Owner: praths71018
Created: 2024-04-19T04:11:11.000Z (over 1 year ago)
Default Branch: main
Last Pushed: 2024-04-24T00:58:21.000Z (over 1 year ago)
Last Synced: 2025-01-29T20:42:22.037Z (9 months ago)
Topics: hindi, indicbert, machine-learning, nlp, product-reviews, sentiment-analysis
Language: Jupyter Notebook
Homepage:
Size: 99.6 KB
Stars: 1
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

README

# Introduction

This project uses different pre-trained models to predict the sentiments of people based on reviews in Hindi.

# Dataset Description

The dataset is a collection of Hindi reviews (2006 training examples) with 4 sentiments:

1. Happy
2. Sad
3. Angry
4. Neutral

# Models used

1. LSTM
2. BERT
3. mBERT
4. DistillBERT
5. IndicBERT

# Steps

## Transliterate

1. In the Transliterate Directory 1st git clone below repository:

```bash
git clone https://github.com/yourusername/torch-transformer-hinglish2hindi-translator.git
cd torch-transformer-hinglish2hindi-translator
```

(go to the repository to check how to use it)

2. Change the model_path in transliterate.ipynb to any of the models and run that file.

## Model building

1. Run Hindi_Sentiment_analysis.ipynb which uses IndicBert model.