https://github.com/vrishinrajpurohit/review-sentiment-analysis
Sentiment analysis web app using BiLSTM on the Amazon Polarity dataset. Built with TensorFlow and Flask for real-time sentiment prediction.
https://github.com/vrishinrajpurohit/review-sentiment-analysis
bilstm deep-learning flask machine-learning nlp sentiment-analysis tensorflow
Last synced: 7 months ago
JSON representation
Sentiment analysis web app using BiLSTM on the Amazon Polarity dataset. Built with TensorFlow and Flask for real-time sentiment prediction.
- Host: GitHub
- URL: https://github.com/vrishinrajpurohit/review-sentiment-analysis
- Owner: vrishinrajpurohit
- Created: 2025-02-28T12:45:24.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2025-03-08T05:48:09.000Z (7 months ago)
- Last Synced: 2025-03-08T06:25:20.354Z (7 months ago)
- Topics: bilstm, deep-learning, flask, machine-learning, nlp, sentiment-analysis, tensorflow
- Language: Jupyter Notebook
- Homepage:
- Size: 0 Bytes
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Review Sentiment Analysis using BiLSTM
This repository contains a deep learning-based **review sentiment analysis** project using a **Bidirectional LSTM (BiLSTM)** model. The model is trained on the **Amazon Polarity Dataset** and predicts whether a given review expresses a **positive** or **negative** sentiment.
## Features
- **Deep Learning Model**: Built using **TensorFlow/Keras** with a BiLSTM architecture for handling sequential text data.
- **Preprocessing**: Includes tokenization, padding, and embedding layers using TensorFlow.
- **Training & Evaluation**: Achieves high accuracy on the Amazon Polarity dataset, demonstrating strong performance in sentiment classification.
- **Inference**: The model predicts the sentiment of new text inputs with confidence scores.
## Dataset
The dataset used in this project is **Amazon Polarity** from Hugging Face:
🔗 [Amazon Polarity Dataset](https://huggingface.co/datasets/amazon_polarity)
This dataset consists of **Amazon product reviews** labeled as either **positive (1)** or **negative (0)**.
## Model Performance
The model was trained for **20 epochs** and achieved the following performance:
- **Training Accuracy**: ~94%
- **Validation Accuracy**: ~93%
- **Precision, Recall, F1-score**: ~0.94
## Web App Screenshot
## Notebook
📌 The complete Jupyter Notebook for this project is available on **Kaggle**:
🔗 [Sentiment Analysis Notebook](https://www.kaggle.com/code/vrishin04/sentiment-analysis/notebook)