Feel free to submit a PR for these ideas. For post similarity, ML learning techniques can be used to calculate the “distance” between two posts, but I don’t know if with an increasing amount of spam could work computation wise. Especially if spammers start using their own GenerativeAI engines.