Fact-Enhanced Synthetic News Generation

The advanced text generation methods have witnessed great success in summarization, language translation, and synthetic news generation. However, these techniques can be abused to generate disinformation fake news. To better understand the potential threats of news, we develop a novel method FACTGEN high-quality content. majority existing either afford limited supplementary information or lose consistency between input output which makes less trustworthy. address issues, retrieves external facts enrich reconstructs claim from generated content improve among output. Experiment results on real-world datasets demonstrate that contents are consistent contain rich facts. We also discuss an effective defending technique identify pieces if was used