I like this layout - in terms of it being straight forward and easy to scroll.
I am curious why some of the stories are trending (for instance (5 sources) at the bottom:
> 1,000 wild horses to be rounded up in Northern California (seattlepi.com)
It appears this is just following / aggregating standard news outlets (I'm guessing ~50 or so). Then clustering the stories based on some sort of fuzzy matching.
Pretty cool IMO, but care to elaborate on the machine learning portion?
If you notice on the page, there are references to remarks made by actual people. The NER feature uses machine learning. Also the continuous clustering employs K-means and not just fuzzy matching as you stated. It is hard to imagine getting anywhere close to the level of accuracy with just fuzzy matching.
2 comments
[ 2.8 ms ] story [ 14.6 ms ] threadI like this layout - in terms of it being straight forward and easy to scroll.
I am curious why some of the stories are trending (for instance (5 sources) at the bottom:
> 1,000 wild horses to be rounded up in Northern California (seattlepi.com)
It appears this is just following / aggregating standard news outlets (I'm guessing ~50 or so). Then clustering the stories based on some sort of fuzzy matching.
Pretty cool IMO, but care to elaborate on the machine learning portion?