Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Collaborative Filtering + Interweb == Good Times?
12 points by earle on May 18, 2008 | hide | past | favorite | 8 comments
Hey folks!

I got a little too frustrated with my RSS subscriptions becoming unmanageable and repetitive. Sites like Techmeme and Technorati have become worthless for me. Digg and Reddit have both broken down tremendously under the extensive growth, and the quality of content as it pertains to me no longer has any value. I don't think joe-shmoe's vote on something should matter as much to me as say, a Hacker News participant reading the same things I do. Submitting URLs for public ranking just seems silly.....

So, decided to get motivated and build a collaborative filtering engine coupled with a feed aggregator that provides recommended, ranked, and categorized content based upon the reading and voting habits of users, as well as article similarities. Users can pick the specific categories of information they are interested in, or define your own.

Since I've basically eliminated the frustrations of my RSS subscriptions through traditional readers, I figured it'd be good to open it up to some users to get some feedback and utilization data for training the collaborative filtering mechanism (linear semantic indexing...).

You can subscribe to a list of fixed sources, or add your own.

Site is up at: http://news.80concepts.com

You can browse ranked and recommended content by categorization: http://news.80concepts.com/category/geekdom/

You can perform time based context queries: http://news.80concepts.com/home/?q=britney+spears&period=weekly

Or even for specific categories/keyword/time based combinations: http://news.80concepts.com/category/entertainment/alltime/?q=twitter

View popular topics via tagged content: http://ofthenow.80concepts.com/tags/

As a nice byproducts: new content source discovery happens naturally and effectively, and searching becomes actually useful for time and trend based queries without the convoluted madness when context is important.

I've also exported similarity data via 2D plot for imported articles via XML as well: http://news.80concepts.com/news-sim.xml I'm sure that could produce some neato-keen-visualization via flash or otherwise if that kind of geekery appeals to someone.

If you would like to see anything specific included, I'm all ears.

We'll take first couple hundred signups....

For those of you who know me, feel free to email me directly... For those who don't (I was, amongst other things, the co-founding CTO of Hotjobs.com) -- I welcome any feedback, complaints, rants, thoughts, or free beer...




I recently joined "Of The Now" and so far I like it. I like that I can sort through all of the titles in one spot rather than having to go to several different sites. I also like that I am basically building my own feed of news that I am interested in rather than what other people think I might be interested in. I look forward to seeing how this site developes.


I stopped hitting slashrot and digg simply because the noise level was unsquelchable. SO off I have been in the land of Greader, my own little feeds set up with some slight input from googles friends linkages...but this...this is cool

Being able to add your own feeds, thats great. I am so sick to death of reading echos of echos of the same hand full of places digg,redit,slash,etc etc pick from. Its threadbare folks. Now i can agg in places like Greylodge and Ovo..whoot.

When I do want to see what the masses are trending for the tools that I see here look to be right on the mark to get the gist and gestalt. Simple and useful.

I look forward to seeing this site build up some more tools and flexibility to bring more diversity , controllable by the user, into the feed agging. One of the problems with how other feed agg sites work is they become so damn insular and inbreed over time. My hope is this one will allow that not to happen.


Like the simple design. The healthcare link isn't too solid at the mo.

http://news.80concepts.com/public/category/science/healthcar...

Why is it that I can't read comments without logging in?


There just wasnt any sources yet for Science -> Healthcare.. I have someone populating the categories as we speak. In the meantime we made it fallback based on similarity and parent categories...

Login decorator for viewing comments is being removed.


Diggin' this project so far.. makes it convenient to catch up on what's happening 'now' on the web..


thanks for the kind words..

that's what it's all about.. and everyones definition of content differs, so we're taking a stab at solving that algorithmically.


Why do you insist on verifying email address? Are you afraid that a bot will join your site and will find other bots with similar interests? :-)

Seriously, what is the threat you are protecting against which makes it worth it doubling your bounce rate?


Validation of e-mails removed :)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: