Please survey the existing methods of text augmentation based on the surface. Please do not be very picky. Just the most common, the one that has been used as baseline the most, ... including their pros and cons, in terms of their applicability in our project (easy coding, also domain)
this paper could be a good start point: EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks