You should check out Feedmarker (www.feedmarker.com) - an RSS/ATOM reader and bookmarks manager that already allows users to tag feeds and bookmarks. Users can not only browse their own feeds using a taxonomy that they've created, but they can also browse all the other users' feeds according to the collective folksonomy. I'd love to know what you think.
Bruno

Hi, Bud. If you're interested in collaborative filtering applied to weblogs, you might check out Findory (http://findory.com ).

It's easy to use. Just read articles, that's it! Findory learns from the articles you read, searches thousands of weblogs, and helps you discover articles you would otherwise miss.

I like both options Bruno and Greg. The one question I have is how would they fit into a corporate setting? As you know, folksonomy and collaborative filtering are highly dependent on the group you have doing them. With both feedmarker and findory, it seems the group is mainly techies.

The defining feature of newsgator in this context is their corporate focus.

Thanks, Bud. I think Findory works well in a corporate setting. Why wouldn't it? And what makes you say that collaborative filtering methods are highly dependent on the group using the system?

At a high level, the algorithms work by finding a sub-community of other users with your interests. Properly implemented, it works well as long as there's at least a few people in the community who share your interests.

Findory is built for a mainstream audience. Hundreds of thousands of readers with a wide variety of interests use and enjoy Findory.

Greg Linden, Founder & CEO, Findory.com

It seems to me that Technorati "tag" searches would be an important place to look regarding these questions.

I would love to find some way to evaluate relative user demographics for the various folksonomy/tagging/social bookmarking tools. I am starting a consultancy for implementation of new web tools, especially for social change organizations and small biz, and would like to know wether there are, for example, more homeschoolers or immigrant rights activists using Delicious, Furl or an alternative. Likewise, what tags are being used would be nice to see. The equiv of Google Suggest, perhaps. (delicious seems to offer something like this) For now, I suppose, there are probably not many users outside of tech, corporate and maybe academic communities of any of these tools, but that's something I hope to help expand.

Greg and Marshall, you raise interesting points. Greg, let me hit yours first:

"Properly implemented, it works well as long as there's at least a few people in the community who share your interests."

That's really the point, whether there are a few people. Now, the question is how many people you need. If I were to adopt a frequentist perspective, my guess would be that I would need a very large number for coefficient estimates to distributed asymptotically normal. Adopting a more Bayesian perspective, I suppose I could have recourse to some notion of prior distribution and get away with many fewer. If I were a computer scientist, I might be satisfied if my classifications from the collaborative filtering alogrithm were stable over time (define stable based on percentage of changing classifications).

Marshall, this snippet strikes me:

"For now, I suppose, there are probably not many users outside of tech, corporate and maybe academic communities of any of these tools, but that's something I hope to help expand."

I thnk you are correct. Have a look at this other post I wrote for IBM's plans in this regard. Folknsonomy is busting loose all over the place:

http://thecommunityengine.com/home/archives/2005/03/ibms_intranet_a.html

I think folksonomies would be just as useful (if not more useful) in a corporate setting as they are in the wild. Corporations are already segmented into various groups that share interests and expertise (technologists, lawyers, marketers, etc.). Allowing those people to categorize content that is relevant to them (and that they know the most about) would help others who don't have the same expertise get up to speed on subjects outside of their field.

For example, if Feedmarker were used in a corporate setting, I (as a PR professional) would be able to easily browse through the feeds that people in the technology department are tagging, thus saving myself the effort of figuring out which technology feeds are worthwhile.

Bruno:

You make a good point. Thanks! I meant my reply comment very narrowly. The narrow idea is:

1. Folksonomy is interesting for aggregating over a group. In some cases, if your group is not represented it could be less useful.

2. If you are going to use various tools like collaborative filtering, you need big numbers.

I agree with your point about people wanting to see how each other is categorizing. You can go even further to note things like when you use tag x, I tend to use tag y; I'd like to see everything you have with tag x. That whole translation can be automated.

See these two posts:

http://thecommunityengine.com/home/archives/2005/04/why_xfolk_fokso.html

where I mention feedmarker as being on the cutting edge, and

http://thecommunityengine.com/home/archives/2005/03/ibms_intranet_a.html

where I discuss IBM's planned and potential use of folksonomy.

Bud