I wrote earlier of Yahoo News Tag Soup, which automatically groups Yahoo News stories into tag categories in a "cloud" format. Want to do the same to your own feed or a
collection of feeds? Yahoo News Tag Soup creator John Herren sends news of his new TagCloud service. Sign-up for free, enter your feed,
and you’ll have your own tag cloud. Or give it a list of feeds, and you can make a cloud in particular subject areas.
I made two clouds to see how it works. The first below shows stories just from the Search Engine Watch Blog. The second collects stories from a range of search blogs I
read. Not all of them are listed, as I didn’t have the time to get everything in there (the ability to take a standard OPML export of feeds would be cool) and some feeds
glitched (sorry, Threadwatch, your feed didn’t validate for TagCloud). No doubt others will in short order setup a comprehensive search cloud including every blog under the sun
(need a list? see this past post). I’ll link
across, when they do.
Yeah, I know, the borders spill across our margins. And technically, I’ve done wrong by putting the link data that should go in the page’s header into the body. But it
seems to work, and trying to put something into one particular post’s header isn’t easy. But you get the idea!
From what I gather, the clouds are based on what’s in a current feed. So I was disappointed in our SEW Blog cloud. We cover a lot of different subjects,
and the cloud doesn’t take that history into account. It would be cool to see what would happen if it analyzed the full-text of all of our posts.
In addition, it remains more fun than useful if you really wanted to drill-down to find stories on a particular topic. We actually categorize all our stories at Search
Engine Watch for members. You can find a list of categories shown on this
page (I’ll be putting a page on the blog listing all of these directly in the future, though
drill-down access will remain a members-only feature).
It would be cool to see that list turned into a cloud based on number of posts. But the underlying data would remain based on us putting things into a set categorization
scheme we have here. That works for us — mileage may vary elsewhere :)