bsd_junkie

16 ปีที่แล้วdiscuss

There are a number of article recommendation engines out there that can fill the need for "outlier" articles fitting even the most peculiar tastes. I personally use http://www.euraeka.com and even though it aggregates news from less hardcore programming sources I find it an incredibly powerful source of science and technology news that fit my taste. I tried Digg and Reddit recommendation engines but they all work on user-to-user based recommendations and most of the time i get either inaccurate or trivial recommendations.

bsd_junkie

17 ปีที่แล้วdiscuss

Hhhm, the link didn't show up: http://www.euraeka.com

bsd_junkie

17 ปีที่แล้วdiscuss

Actually, blindly submitting ALL stories from those sources is not the smartest thing to do. Number one, you flood the system. Number two, there is no guarantee that new articles from those sources will necessarily do well. The more refined way to do this is to train a recommendation engine from an article aggregation system on some of the top articles that have been most liked on HN. Then, as the engine will collect and rank articles for you automatically and you can just use your script to take the article from the recommendation site and post them to HN. This way you have more assurance that your submissions will be liked on HN, and also the engine will likely find new article sources that you have never heard of before. I friend of mine is using this method to become a power user on Digg and Reddit. I think the uses several sites and has coded his own ranking system but the bulk of his recommendations come from www.euraeka.com, which is a news aggregator and recommendation engine. Basically he signs up for an account, finds out what the top articles on Digg are for lets say last year, then picks the top 25, searches them in Euraeka and then the engine starts finding articles that are like those. All he does afterwards is submit the recommendations on Digg. I think he does the same for Reddit. Anyways, Euraeka is just one of the engines out there. I am sure you can find others as well. The only question is which engine is spitting out the "best" recommendations. By "best" I mean which engine manages to most closely estimate the taste of the crowds on sites like HN, Digg, Reddit, etc.

bsd_junkie

17 ปีที่แล้วdiscuss

Actually, any service out there that tries to do away with the requirement that people actively perform some effort would fit the bill for a potential Google killer in a particular vertical. In other words, any recommendation service that preferably can "learn" the habits of its users and then simply recommend new content. DirectedEdge is just the latest player in this game. I personally track the news recommendation niche. There used to be a service called Findory that in my opinion came very close to being "optimal". But they went under for some reason. The closest to Findory (and with some very cool features such as computational lie detector for news) is http://www.euraeka.com They have all three basic features - search (aka Google-style), discovery (aka crowd-wisdom style) and recommendation engine.

bsd_junkie

17 ปีที่แล้วdiscuss

I think a site that was submitted here about a month ago has almost the same features except that they fetch news from everywhere, not just HN and order them based on several interesting scores like "popular", "controversial", etc. They call it <a href="http://www.euraeka.com>All the news that's fit to read</a>. Pretty bold claim:-) From what I can see on their site, when you register you can "save" your sorting preferences (click on Advanced at the top of the site) and the site also allows you to filter by topic and domain. I am personally using it for their recommendations system. Truth be told, all news sites are really just a portal so you can find what's interesting to you. Once a system is trained on recommending you what you like, then all the sorting, scoring, mashing up, etc becomes pointless. That's goal really of every information providing site - customize the information flow to every user to the point that every user sees a unique version of the site based on their unique preferences. So, while I like the sortable HN concept ultimately it needs to be extended to a personalized recommendation news system in order to be truly useful. Euraeka and Google News are two sites that I like, Digg's recommendations frankly suck in my opinion. Good luck to you though, good stuff.

comments