Comment by balksi
I am working on yet another news aggregator: newsmuncher.com
So far i've got the scraping and embeddings / similarity clustering down (to build timelines of news stories), lots of data cleaning and UI refinement required. I find it hard to make choices, maybe I need a cofounder who can pair up with me. Looking to either monetize news data or build a news analysis / intelligence platform.
May I ask what techniques either you're using or would recommend for similarity clustering? I looked into topic modeling, but it seemed a long way off from reliably bundling together stories like on Techmeme.
(I'm working on basic blog and video aggregators like Planet Python.)