There’s no need for a cut-off date; just take highly upvoted comments off Reddit and find similar comments on HN as determined by word vectorization, and then t

There’s no need for a cut-off date; just take highly upvoted comments off Reddit and find similar comments on HN as determined by word vectorization, and then train a model off of that. Or maybe even skip the similarity step; after all the goal is to find Reddit like comments.