New RSS filtering site FeedScrub launches today in private beta. 500 SitePoint readers can join the beta by using the invite code ‘sitepoint‘
FeedScrub, which is initially targeting bloggers and other high volume consumers of RSS, uses an adaptive filter that learns as you read and filters RSS feeds based on your input. Many current RSS filters use keyword lists to filter out (or in) content, but FeedScrub works more like a spam filter, learning about the content that you do like, and removing content that you’ve indicated you don’t like.
“Existing RSS filters focus on filtering out particular keywords, creating the possibility of false positives (deleting something the user wanted),” says founder Jason Ardell. “[FeedScrub] is personalized to your preferences so you only read what interests you, and it gets smarter as you train it to minimize filtering out stories you’d like to read.”
The site currently supports Google Reader and the NewsGator RSS readers (NewNewsWire and FeedDemon), with more RSS readers on the way. (Any reader can be used by accessing the raw filtered feed URL.)
The video below, which comes from FeedScrub, explains how to use the application with Google Reader.
The way FeedScrub works is very simple. You add feeds to the site, which creates special scrubbed versions. You then add the scrubbed feeds to your RSS reader and start using buttons included with each entry to mark the posts you like, and the ones you don’t. Eventually, FeedScrub’s algorithm will learn what type of content you like and what type you don’t, and begin to filter your feeds accordingly. The site also includes a junk feed so that you can keep an eye on the stories that are getting culled in case something you like gets scrubbed. That’s especially useful while the algorithm is still being trained.
Personally, my problem with this type of filter is that I often come across articles in my feed reader that I really enjoy about topics that I wouldn’t normally want to read. Often times these comes via feeds I subscribe to from social news sites like Reddit or Digg. I may not want to read every article that comes through about the auto industry, for example, but once in awhile one that was flagged by my peers as interesting will catch my eye. Further, because I generally only read feeds from blogs whose content I really like, I end up tagging too many posts as “I like this!” for the filter to learn anything helpful about what I’ll want to read.
That’s why PostRank (our coverage), which filters feeds based on audience engagement might be a better model. Chances are, if you’re subscribing to a feed you already like most of the content, but if you’re overloaded you just want the best of that content to bubble up.
That said, you could certainly use the two filters in tandem. Run a feed through PostRank, then again through FeedScrub. That way you’ll get the top posts from your favorite blogs and filter out anything that the crowd liked but you don’t.
Remember that we have 500 invites to FeedScrub for SitePoint readers — just use the invite code ‘sitepoint‘ when signing up. Let us know in the comments what you think of the service.
Josh Catone joined Mashable in May 2009 and is Executive Director of Editorial Projects. Before joining Mashable, Josh was the Lead Writer at ReadWriteWeb, the Lead Blogger at SitePoint, and the Community Evangelist at DandyID.