For Usenet, you could filter posts based on scoring of posts based using regexps on the various fields of a message - body, subject, sender, etc. See the following for some details:

http://www.slrn.org/docs/slrn-FAQ-4.html

I meant PageRank as just one example of Collaborative Filtering, where Amazon and NetFlix collaborative recommendations would be another, and probably more relevant. I think we could "like" someone on the email list, and their "likes" would count transitively to us, and the likes of those they like, etc.