Sentiment Analysis GRU Model

This project implements a Sentiment Analysis model using a neural network architecture built with Keras. The model classifies text into binary sentiment categories (positive or negative) by leveraging word embeddings and Gated Recurrent Units (GRU) layers. Additionally, the project includes a simple Flask web application to provide a real-time sentiment analysis interface.

Installation

Clone the repository and install the necessary dependencies using pip.

git clone https://github.com/QuocAnh189/Sentiment_Analysis_EventHub
pip install -r requirements.txt (note python 3.8.10)

Key Features

Tokenizer Initialization

• A Tokenizer object is initialized to handle the text data, limited to the 10,000 most frequent words.
• The tokenizer converts the text into integer sequences using fit_on_texts(X) to prepare the input for the model.

Model Architecture

The model leverages three GRU layers for its architecture:

Embedding Layer

• Converts word indices into dense vectors with an embedding size of 20.
• Handles an input dimension of 20,000 to accommodate the vocabulary.

GRU Layers

• First GRU Layer: Contains 16 units and returns sequences.
• Second GRU Layer: Contains 8 units and returns sequences.
• Third GRU Layer: Contains 4 units and does not return sequences.

Dense Layer

• A fully connected Dense layer with 1 output neuron using a sigmoid activation function for binary classification (positive/negative sentiment).

Model Compilation

• Loss Function: Binary cross-entropy (binary_crossentropy), as it's a binary classification task.
• Optimizer: Adam optimizer with a learning rate of 0.001.
• Metrics: The model is evaluated using accuracy.

Usage

To use the Flask web application for real-time sentiment analysis:

Running the Web Application

1. Start the Flask development server:

   flask run or python app.py

2. Open your web browser and go to http://127.0.0.1:5000/. You'll see the home page where you can input text for classification.

3. Enter the text and submit it to get the sentiment prediction.

Running with Docker

In project terminal run docker-compose up --build Test api in Postman with BASE_URL and GET METHOD: http://127.0.0.1:5000/predict

Model Prediction Flow

• Tokenization & Padding: Input text is tokenized using the pre-trained tokenizer and padded to match the expected input length of 241 tokens.
• Prediction Output: The model outputs a probability between 0 and 1. If the value is greater than 0.5, the sentiment is classified as Positive, otherwise it's classified as Negative.

Example

Here’s a quick example to demonstrate how the sentiment classification works:

1. Input: "I love this event, it is interesting"

   • Output: Positive (e.g., 0.85 probability)

2. Input: "The event was terrible and very disappointing."

   • Output: Negative (e.g., 0.15 probability)

Author Contact

Contact me with any questions!

Email: [email protected] Facebook: https://www.facebook.com/tranphuocanhquoc2003

Thank You so much for your time !!!