GitHubHelp

NarendraYSF/LSTMEmotion-Sentiment-Analytics

An LSTM-based sentiment analysis model for classifying text emotions. Built with deep learning techniques to accurately detect and predict sentiment in text data.

★ 22Forks 0Jupyter NotebookGitHub ↗Compare
aiemotion-detection-emotion-classificationjupyter-notebookkeraslstm-sentiment-analysisrecurrent-neural-networkssocial-media-analysistensorflow

README

Sentiment Analysis on Movie Reviews

Project Overview

This project implements a sentiment analysis model using Long Short-Term Memory (LSTM) networks to classify IMDB movie reviews as positive or negative. The model leverages deep learning techniques for text analysis, providing a robust solution to evaluate user sentiments.

Features

  • Binary classification of movie reviews (positive or negative).
  • Preprocessing pipeline with tokenization and padding for textual data.
  • LSTM-based architecture for sequential data learning.
  • A user-friendly function for real-time sentiment predictions.

Technologies Used

  • Programming Language: Python
  • Libraries:
    • TensorFlow/Keras
    • Pandas
    • Scikit-learn
  • Tools: Kaggle API for dataset retrieval

Dataset

  • Source: IMDB Dataset of 50K Movie Reviews
  • Access: Downloaded via Kaggle API.
  • Structure: Includes 50,000 movie reviews labeled as "positive" or "negative."

Workflow

1. Dataset Handling

  • Download: Use the Kaggle API to fetch the dataset.
  • Extraction: Extract the CSV file from the zip archive.

2. Data Preprocessing

  • Load the dataset using Pandas.
  • Convert sentiment labels into numerical values (positive: 1, negative: 0).

3. Train-Test Split

  • Split the data into 80% training and 20% testing subsets.

4. Tokenization and Padding

  • Tokenize text to convert words into sequences of integers.
  • Apply padding to ensure consistent sequence lengths for the LSTM model.

5. Model Architecture

The LSTM model consists of:

  • Embedding Layer: Converts word indices to dense vectors.
  • LSTM Layer: Processes sequential data for sentiment classification.
  • Dense Output Layer: Sigmoid activation for binary classification.

6. Model Training

  • Trained for 5 epochs with a batch size of 64.
  • Validation split: 20% of the training data.

7. Model Evaluation

  • Evaluate the trained model on the test dataset.
  • Metrics: Accuracy and binary cross-entropy loss.

8. Sentiment Prediction

  • Implement a function to classify the sentiment of user-provided reviews.

Results

  • The LSTM model achieved significant accuracy on the test data.
  • Example Predictions:
    • Input: "This movie was not so interesting." -> Prediction: Negative
    • Input: "This movie was very amazing." -> Prediction: Positive

How to Run

  1. Clone this repository:

    git clone https://github.com/NarendraYSF/LSTMEmotion-Sentiment-Analytics.git

  2. Install the required dependencies:

    pip install -r requirements.txt

  3. Run the Jupyter Notebook to train, evaluate, and test the model.

Future Scope

  • Expand Dataset: Include more diverse reviews to improve generalization.
  • Optimize Hyperparameters: Experiment with different learning rates, batch sizes, and epoch counts.
  • Alternative Architectures: Explore GRU, Transformer-based models (e.g., BERT).
  • Deploy: Build a web or API interface for real-time predictions.

License

This project is licensed under the GNU General Public License.