Abstract:
"The dissemination of false information on social media has significantly increased during
the past few years. The purpose of this study is to suggest an innovative method for
identifying incorrect news utilizing data from Twitter. Two methods were used, one based
on text data-based prediction and the other on image-based prediction. The tweets in the
dataset were all classified as either false or authentic. For text-based detection, ten machine learning models were created, including Decision Tree, KNeighbors, Random Forest, and Logistic Regression. Additionally, developed to enhance classification performance was an ensemble model. For image-based detection, a deep learning Convolutional Neural Network (CNN) was developed to distinguish between real and fake images associated with news articles.
Users can now interact with machine learning models and get predictions due to a GUI.
The models then provided the expected classification once the user had entered the required variables of the tweet or an image associated in the news into the GUI. In addition, a Power BI dashboard was developed to show the hidden insights and patterns of false and authentic news items graphically.
The outcomes demonstrated that the machine learning models could classify tweets and
photos as real or fraudulent with great accuracy. The ensemble model, which had an
accuracy of 89%, performed the best. The classification accuracy of the CNN model for
images was 73%.
In conclusion, the suggested method showed how machine learning and deep learning
models performed well in identifying fake news using Twitter data. The models were easier
to access and understand due to the GUI and Power BI dashboard. This approach offers a
possible direction for detecting fake news studies in the future."
