Imagine this

Python + Rewriting proxy (multiple out there, i know... maybe using Amara) + Bayesian filtering.

Assume the spam filtering would be on a per div and/or span basis...

Indeed, just think about it... Wouldn't this (after being trained):

• ignore popups, probabely because the popup text is blocked, but this could be on a per window/file basis of course


• strip text only ads most other software can't strip


• strip commercial images just like many other application would (like the marvelous AdBlock firefox extension), but now with a learning touch to it?


• integrate with the browser if xul would be used?

I'd love to imagine this happening, on my machine..
Anyone interested into building something like this?