Our Hybrid lda2vec Algorithm
multithreaded.stitchfix.com
multithreaded.stitchfix.com
I also like the honesty of this report, mentioning different methods that are similar (even another project called lda2vec) and the disclaimer that this is probably not for everyone. The Python code does make it more accessible however, so I could see myself at least reusing concepts that are implemented here. On the whole this looks like an interesting algorithm for the stated purposes.
I guess that was written in gest, but I do hope some kind of reddit or SO-style siloing isn't on the cards to 'make HN grow'!
I rather hope that HN grows in the collaborative filtering direction instead. Instead of most interesting topics never making it high enough up the first page to actually gain any traction, it would be nice if collaborative filtering was used to make it so people can easily get more of what they like and easily find something completely different. Of course I've blogged and petitioned for this kind of approach in the past ;) http://williamedwardscoder.tumblr.com/post/15581427232/self-...
Then HN can stay the HN we all like, as many of us like slightly different things, and collaborative filtering can evolve with each of us letting us find like-minded others to share our filter-bubble with :)
If HN were profit-optimizing and let anyone put ads, it would break the site completely as the density of ads makes it more difficult to read content (something many many blogs like Forbes fail to realize).