Which is the best sentence-embedding technique for the following use case?

I want to get some advice regarding how I should go about the following problem. Given 5 labels and 5 sentences I want to match each sentence to the most appropriate label. I believe that sentence embedding is the best technique for this, but these labels and sentences will be containing a lot of technical jargon which isn't as common in the pre-trained datasets. I have a collection of textbooks that I would like to use for transfer learning so …

case embedding languagetechnology sentence-embedding