Train based on text in body

Thank you!

One question - is it possible to set it such that I archive only some of the sources, or at least some folders of sources?
In my case, I have many sources (I should prune at some point), and it’d be wasteful & excessive to archive everything from all those feeds.