This is totally cool and kudos to you. But be aware there are limits of personalized hacker news. I think Bill Maher put it best when ranting just last night about facebooks customized news feeds:<p><i>Newspapers may be old-fashioned, but here's what we're losing if you never see one. They are trying to tell you what's actually important, not just what's important to you. You may not read the whole paper, but you at least see headlines, making you aware that something's going on outside of your microtargeted world of fashion or music or Wiccans or zombies or whatever you're into.</i><p>Replace 'newspapers' with hacker news and you get the point.<p><a href="https://www.youtube.com/watch?v=WohtmZDZCGM" rel="nofollow">https://www.youtube.com/watch?v=WohtmZDZCGM</a>
"The model can only get better with more training data, which requires me to judge whether I like stories or not. I do this occasionally when there’s nothing interesting on Facebook. Right now this is just the above command-line tool, but maybe I’ll come up with something better in the future."<p>If you let your program log into HN using your account, it should be able to tell which of the stories you've up-voted there. If you use that as the input to your classifier, as you read stories on HN, simply mark those that interest you by up-voting them.<p>I'm also curious to know whether the stories are weighted by age to account for changes in what you find interesting.
FYI, the new Hacker News API allows easy programmatic access of story/comments and infinite chronological paging. You could download <i>every</i> Hacker News story in less than 2 hours without breaking the API request limit.<p><a href="https://hn.algolia.com/api" rel="nofollow">https://hn.algolia.com/api</a>
If this is a solution to the "not enough links that I personally like" - kudos to the author. Nice to find a fun project to work on that will also solve a problem for them.<p>I personally despise recommendation / personalization algorithms of any kind. I still have never found one that's actually better than myself at distinguishing articles that I'd like to read, music that I want to listen to, tweets I'd like to see, etc.<p>When reading HN, I'm constantly surprised by links that would not normally be on my radar for things I'm interested in. I think personalization algos, in general, are good at filtering those away.<p>Since the author mentioned HN being too much of a firehose and this then also being a solution to the "too many links to keep up to date on" problem, the solution might be a bit simpler than the author suggested.<p>HN already has the "best" links at <a href="https://news.ycombinator.com/best" rel="nofollow">https://news.ycombinator.com/best</a><p>It's hard to find - it's in the 'Lists' section in the footer, but it's still there and I use it all the time, when I haven't been actively reading HN for a while.
ideally you could train the data set from stories i've upvoted on HN<p><a href="https://news.ycombinator.com/saved?id={{username}" rel="nofollow">https://news.ycombinator.com/saved?id={{username}</a>}
It seems you prefer to read articles from:<p>- WashingtonPost<p>- BusinessWeek<p>- MarginalRevolution<p>- NY Times<p>[1] Based on BuzzSumo's social data:<p><a href="http://app.buzzsumo.com/#/influencers?q=@joelgrus&type=influencers&result_type=relevancy&blogger&influencer&company&journalist&regular_people&ignore_broadcasters=false&page=1" rel="nofollow">http://app.buzzsumo.com/#/influencers?q=@joelgrus&type=influ...</a> (Press View Links Shared, Analyze Links Tab]
It seems there is a small but very strong subculture of HackerNews readers who enjoy reading and discussing mathematical things. I would love to have a separate feed of those stories (and then after I'm done I could browse the HN front page), and I have often thought about the possibility of writing a program to do that.<p>DataTau (the HN for data mining) seems to have failed, so I imagine a filter is the way to go rather than make a new website.
I just use the 50 or 100 point minimum feed in my reader, and skip articles that don't look interesting based on how much time I want to spend. Sometimes I only read articles if they're a day old and the first comment makes them look interesting.
Cool story bro. You may want to check out digitalocean for hosting - their cheapest option is only $5 a month - about equivalent to the smaller $40/month aws option. It's very simple as well.