JohnyDL
commented
5 years ago I think this is a really cool idea
I was gonna make my own submission but I thought it better to bump this one,
Maybe A Browser Extension that maybe follows you around the web? It could read pages that are linked to ones you visit (maybe with depth being an option) then gives them a rating for that/those link(s) based on how similar that page is to pages you've been to in the past and liked (I guess this would work best on pages with built-in like buttons), or specifically to how similar it is/they are to the page you're on right now (which might be useful if you're looking for news/information on a given topic). It could even recommend pages to you that're down the rabbit hole so to speak.
A more niche use case might be a Youtube Crawler Extension that visits videos you visit and puts them through an NN to analyse them and trains for how likely they are to earn your thumbs up or thumbs down (maybe it could optionally parse the video and audio (probably audio is the more important of the pair as music and tone might be interpretable but visual styles aren't so easy to quantify) as well as the text) and then it visits any pages youtube default recommends to you/links on the pages you visit (even outside youtube)/in your sub box and gives you a % for what it thinks the chances are of you liking the video vs disliking the video maybe even your chance of commenting or not doing any of the above too. And because it's a browser extension and all done on your own device you own all that data about you and you get to decide when it's no longer relevant rather than relying on google's one size fits all.
This is a suggestion for a future topic. Maybe more of a python tutorial, I don't know if it would make sense to use Processing or P5 for this.