Data Augmentation for Fake News Using ChatGPT

With the development of the Internet, the adverse effects of fake news on society have become increasingly serious. The use of machine learning techniques, such as natural language processing and deep learning models, to automatically identify news and prevent the spread of fake news has become a good solution. However, due to privacy concerns, fake news is not easily obtainable, resulting in a scarcity of news data. To address these issues, we utilize ChatGPT to generate news data (ChatGPT-based data) through designing high scalability prompt. By comparing the distribution of ChatGPT-based data with the original news data, we validate the effectiveness of our method. We design a simple neural network classification model and compare its performance when using and not using the ChatGPT-based data. Experimental results demonstrate that the ChatGPT-based data is usable and high quality, providing an effective solution to the scarcity of data for fake news detection.