Hello! So I have a corpus of text that has very domain specific words and meanings. For example, it contains products, their descriptions, keywords, commonly used words to refer to the product, etc. Now, I want to utilize a Q&A model for another task but I would like for it to understand these domain terms. The problem is that I have no Q&A training data available for my domain and to create this data for the corpus would take ages. I was thinking that I could train a MLM on this corpus of text and then transfer what that model has learned about the domain and apply it to a Q&A model.
Is this possible? If yes, how should I proceed?
Thanks!