My dream is to build a browser to do this for the entire web, using AI/ML to "translate" ad-ridden sites to their pure HTML/CSS counterpart. Lots of pre-trained models can make sense of pages based on visual layout. Seq2seq/summarizer models can work hand-in-hand to take out the cruft.
I imagine a cache of this simple web shared across all browsers.
Imagine translating all PDF manuals, scientific research, and yes, recipes, in this decentralized manner. The dataset would be clean both for human consumption as well as ML training.
I imagine a cache of this simple web shared across all browsers.
Imagine translating all PDF manuals, scientific research, and yes, recipes, in this decentralized manner. The dataset would be clean both for human consumption as well as ML training.