Sunday, December 7, 2008

Feedvis generates a tag cloud and lists word frequencies from an opml export of your rss feeds. Below, for example, is a section of the feeds in the renowned Knowbodies/Netvibes metagator. On the left, the words and their frequencies, and on the right the blogs issuing the feeds. You can analyse all feeds collectively (as I've done below), or analyze one particular feed. If you're interested in tag clouds (personally I think they look nice, but don't understand what they're good for) is a neat plaything. Regarding the permanence of the site, the person who developed it says "Not very, it's mostly a project to help me learn programming."


  1. Hi, thanks for mentioning my FeedVis program! I wanted to let you know that, since FeedVis has been so well-received, I'll be supporting it at least through 2009.

    You raise a good point about the usefulness (or lack thereof) of tagclouds; I've posted about that very topic. There's been interesting research (see my post for citations) suggesting that tagclouds may be less useful than simple ordered lists for some applications. Tagclouds also are deceptive when it comes to longer words (they look more frequent than they are) or words with lots of descenders and ascenders (the vertical lines like on 'b' and 'p'). Tags made from word frequency rather than tag frequency have further problems; it's less than clear that simple measures of frequency are the best way of getting the gist of a text. Matthew Hurst raises this good point in his recent post about FeedVis.

    This project was actually an attempt to alleviate one of tagclouds' main flaws, the lack of chronological context; I think it does a fair--not great--job of that. but there's no doubt that there's quite a bit of improvement left to do. I'm always open to suggestions!

  2. oops, here's the link to the post I mentioned by Matthew Hurst: