AG News Text Classification

This repository contains code for text classification on the AG News dataset using various machine learning and deep learning techniques. The AG News dataset is a collection of news articles categorized into four classes: World, Sports, Business, and Science/Technology.

Result

Dataset

The AG News dataset consists of 120,000 training samples and 7,600 test samples, evenly distributed among the four classes. Each sample is a short news article accompanied by a title.

You can download the dataset from here.

Dependencies

• Python
• NumPy
• Pandas
• Scikit-learn
• TensorFlow (for deep learning models)

Models

This repository includes implementations of the following models:

• Multinomial Naive Bayes
• SGD (Stochastic Gradient Descent)
• Convolutional Neural Network (CNN)

Usage

All the necessary steps for data preparation, preprocessing, and model training are contained within the Jupyter Notebook provided.

1. Data Preparation: Download the AG News dataset and place it in the data directory within the same directory as the notebook. If you're using a different dataset, make sure it follows the same format or preprocess accordingly.

2. Preprocessing: Execute the preprocessing cells in the notebook. These cells will tokenize the text, remove stopwords, and convert the labels into numerical format.

3. Model Training (Optional): If you want to train your own models, execute the training cells in the notebook. You can use the provided scripts or your custom implementations to train the models.

   • Ensure that you have the necessary dependencies installed in your Jupyter environment.

4. Prediction App: Use the provided prediction app to classify news articles. Here's how to run the app:

   • Ensure that you have the necessary dependencies installed.

   • Run the prediction app using the following command:

     python app.py
     

   • The app will prompt you to input the text of a news article. Enter the text and press enter.

   • The app will then predict the category of the news article based on the trained model.

5. Evaluation: If you want to evaluate the performance of the trained models, execute the evaluation cells in the notebook. This will output accuracy and other evaluation metrics.

6. Demo

Contributing

Contributions are welcome! If you have ideas for improvements or new features, feel free to open an issue or submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.