Sentiment Analysis with Deep Learning

This project is a sentiment analysis model that uses deep learning techniques to predict the sentiment (positive or negative) of movie reviews. It's based on the IMDB movie review dataset, containing 50,000 movie reviews.

Project Overview

Sentiment analysis, also known as opinion mining, is a natural language processing (NLP) technique that determines the sentiment or emotion expressed in text data. In this project, I've built a deep-learning model to classify movie reviews as positive or negative.

Dataset

The dataset used in this project is the IMDB Movie Reviews dataset from Kaggle. It consists of 50,000 movie reviews labeled as "positive" or "negative."

Model Architecture

• Word Embedding: GloVe word embeddings are utilized to convert words into numerical vectors.
• LSTM Layers: Long Short-Term Memory (LSTM) layers capture sequential dependencies in the text data.
• Spatial Dropout: SpatialDropout1D is applied to prevent overfitting.
• Dense Layer: A fully connected layer with sigmoid activation produces sentiment predictions.

Project Components

1. Data Preprocessing: I cleaned and preprocessed the text data, including removing duplicates and encoding labels.

2. Word Embeddings: I used pre-trained word embeddings (GloVe) to convert words into numerical vectors to capture semantic meanings.

3. Deep Learning Model: I built a deep learning model with an embedding layer, LSTM layer, and dense layer for sentiment classification.

4. Training: The model was trained on a portion of the dataset and evaluated on a test set.

5. Inference: I provided examples of how to use the trained model to predict sentiment on new movie reviews.

Dependencies

Make sure you have the following Python libraries installed:

• TensorFlow
• Pandas
• NumPy

Usage

• You can train the model by running the provided or Python script.
• To use the model for prediction, follow the example provided in the script, where you can input your own movie reviews.

Model Saving

The trained sentiment analysis model is saved as sentiment_analysis_model.h5. You can load this model to make predictions on new data without retraining.

License

This project is licensed under the MIT License. Feel free to use, modify, and distribute the code.

Acknowledgments

• The GloVe word embeddings used in this project are available at https://nlp.stanford.edu/projects/glove/.
• The IMDB Movie Reviews dataset is available on Kaggle.

For more details and code, please refer to the project files.