Facebook Plunges Again
Tumblr 101: Tips And Tricks To Get Started
Feedhub: Helping reduce RSS overload?
Starting off with your own RSS feeds (which you upload in an OPML file from your favoured newsreader), FeedHub analyses the content in that file, determining content you seem to be interested in (grouping them into memes). Whilst the initial analysis is pretty smart, from there on you do need to help rate and dismiss memes and individual posts for relevancy to hone the system. This is done both via the FeedHub website, and within the newsreader as FeedHub inserts a relevancy 'flare' into each post.
Our initial guidance to the system made a sample selection of daily feeds go from 235 feeds (and roughly 1,000 posts) to just a dozen or so posts. After a little tweaking, the amount of content increased and still remained relevant. Whilst in theory and practice a fantastic idea, our main concern is that people just don't have enough time, or rather inclination, to train FeedHub - Robert Scoble also raises some interesting questions (and interviews the folks at FeedHub) on his well-respected blog. If you're in need of trimming RSS your RSS feeds and saving yourself time, FeedHub might be of interest - even if like us, you resort back to simply unsubscribing from feeds instead.
Comments
4
Subscribe to commentsPeterSep 30th 2007 2:47PM
The problem with all these recommendation systems is that they create positive feedback loops which keep narrowing what you are exposed to.
If I like baseball and it sees I have some baseball feeds, it will give preference to baseball related items. Since those items are pushed to the top, I read those more. Then it thinks I am REALLY interested in baseball, so it offers more baseball items and pushes everything else to the bottom. Eventually the only things left are baseball related and everything else has been removed.
You would have to work hard to keep an array of unrelated feeds balanced in terms of what it thinks your level of interest in them is.
Sean AmmiratiSep 30th 2007 3:22PM
Hi, I'm the VP of Product Management and Business Development at mSpoke.
Thank you for the review! So you know, we also can learn about you from other elements of your digital identity such as your link blog, delicious or digg account. (See - http://www.mspoke.com/blog/?p=6)
Thanks!
- Sean
Sean AmmiratiSep 30th 2007 4:49PM
Peter -
We have a nuber of techniques to address the issue you discuss. One of them is to also have memes about what is popular.
For example, we look at topics that are popular & the number of incoming links to a post. We also pull data from wisdom of crowd web services - specifically memes that recommend items from the Delicious hotlist and Digg's most popular feed.
- Sean
Fred ThompsonOct 1st 2007 12:02AM
What a whiner!
If you don't have the time to tell the software your relevancy opinions, you don't have the time to put into RSS.
Q: How is this any different than telling a spam filter relevancy opinions?
A: It's not any different.