In this project, the process of data annotation for fake news in the Hindi language is done. Data annotation is a very important task and essential for any AI/ML/DL project. Annotated data is used for training AI/ML/DL models, models learn for this training data and then work on the test data. Although the task of data annotation when done manually is very tedious and time-consuming it plays a huge role in determining the accuracy of the model. Data annotation is the key to building a successful AI model with high accuracy. The higher the accuracy is the better are the results of the model. The accuracy of the model depends greatly on the quality of the annotated data. If there is even a slight inaccuracy in data annotation the overall accuracy of the entire model is greatly affected.

A machine-learning and a deep learning based framework to automate the process of data annotation. Our main contributions are:

The proposed models are tested on 10%, 20%, 30% and 40% test data of the dataset prepared. Our model has shown very promising results with high accuracy of 81.44% for the Random Forest model implemented on 10% test data. The highest accuracy for the LSTM model having 100 epochs and a batch size of 64 implemented on 10% test data was 64.70%.

Results of ML and DL Models