With increasing public access to social media, many dubious and inaccurate content is being generated and shared for profitable targets. This content is generated to attract audiences, increase revenue, impact people’s decisions, and influence important events such as political elections. Manually, detecting this news is very time-consuming, costly, and tedious, so the automatic detection of this content has attracted the attention of many natural language processing researchers. In this research, we have investigated the effect of using EDA data augmentation methods and the ParsBERT pre-training model to solve the problem of scarcity and lack of annotated datasets in the Persian stance detection task The results of this study indicate that we can identify the stance of news to the specific claim better than previous works with the help of data augmentation methods, content-based representation of the data, and the ParsBERT model.