When I saw this title I assumed it was about using ML to build dynamic scrapers that self-configure where the changing data elements. Does that exist anywhere?

I think this is what Diffbot does !

At a very high level, it's similar. We use computer vision and ML to extract structured data from any web page, even ones we haven't seen before. https://www.diffbot.com/

If anyone has any questions or wants to try it out, feel free to email me directly at dru@diffbot.com

interesting read. I liked the part about identifying rare templates. plus, if I ever plan to do ML on html the article gave me a good place to start on feature extraction.

might have some uses in identifying phishing urls, etc

The title should be "Clustering HTML pages". It's not a terribly interesting application. The only thing new I got from it was a de-noising technique.

Not sure why you're getting down-voted. You're right, the author didn't close the loop. Typically we would expect to see an insight after you apply the ML technique.

So they extracted features, clustered the pages and found... what?

I am sure they learned something but it might be proprietary.

What I'd really like to see is a presentation of an algorithm that automatically recognizes and hides the first dismissive HN comment that inevitably appears. Any takers?

Recognize, hide, and automatically post to /iamverysmart.

You can do it yourself, if you're so inclined. Out-of-the-box sentiment analysis is ~90% accurate. Feel free to provide training data for it.

Recursively?

and the title of the browser tab is flashing at me...