Hi,
I’ve been thinking about how Pi-hole can be improved and just thought that something, to help with server side ad injection technologies. I’ve landed on considering implementing a hybrid, blocklists and on device (or cloud based) ML blocking option. The blocklists would be a fast option blocking most ads, but then those which are more evasive would then be handed off to a ML model, to classify the content so it can be blocked or not.
Would this help with catching more advertisements, so can be more likely to catch and stop scam advertisements and any still lingering malvertising campaigns? Is this implementable for the open source project? What do you guys think please?
Thank you for your time and attention in this matter, as well as all of the work on the platform and underlying software.
Regards,
John