Agree with you on this. For text data, there was a paper called "UDA"(https://arxiv.org/abs/1904.12848 https://arxiv.org/abs/1904.12848) that did some work on t

Agree with you on this. For text data, there was a paper called "UDA"(https://arxiv.org/abs/1904.12848 https://arxiv.org/abs/1904.12848) that did some work on this direction. They augmented text by using backtranslation. Basic idea is you take text in English, translate that to some other language say French and then translate back the French text to English. Usually, you get back an English sentence that is different that the original English sentence but has the same meaning. Another approach they use to augment is to randomly replace stopword/low tf-idf(intuitively say very frequent words like a, an, the) with random words. You will find implementation of UDA on GitHub and try that out. I am learning these existing image semi-supervised technique right now and the plan is to do research on how we can transfer those ideas to text data. Let's see how it goes.