I’ve used Flickr’s folksonomy for fun, randomly touring tags, and once for work to find creative commons portraits i could use for persona. SO that worked out for me.

I use the del.icio.us folksonomy a lot to find links. I got interested in a Nintendo DS Lite, but Google was feeding me ads and useless links, but delicious came through with useful links. I wonder if it will fall prey to marketers and spammers eventually, though, the way meta keywords did back in the day.

After reading yesterday’s article, I was wondering when this one would appear. Did not have to wait long ^_^ In my opinion, folksonomies are better for scale but are more difficult to cognitively map. With a taxonomy you have a line that can be drawn from your start position to the end position following the paths of decisions in the taxonomy structure. Folksonomies tend to lack a structure which can make them more difficult to use when searching for something particular. They rely too heavily on user input to make them functional. Places like digg work because so many people are providing input that eventually data is tagged in a useful way.

Like David, I was waiting for some examples of scalable Folksonomies after the post yesterday. And while, like Ben, many of us have used folksonomies successfully to browse, or for fun, where ARE the examples of large, well-scaled ones? I must disagree, at leasat a little, with David’s comment that in folksonomies things can occasionaly hard to find. I used a tagging system for my home photo collection, and at 5k items, I consider it mid-sized. I can find photos fairly easily thanks to cross referencing multiple tags. However, I still have that mental hangup where I feel they aren’t quite as organized as they could be. I guess the key is the correct use of tags, in a way that is intuitive. Using the old reporter’s axiom of the 5 W’s – Who, What, Where, When, Why – gives you a tagging system that is, I believe, fairly sound, and makes things quite easy to find. However, in a social network, things quickly dissolve, as you have hundreds or thousands of people picking tags, each with a different mindset of what is important to tag.

I think folksonomies are quite usable as they scale. I use them on my blog (small scale); I’m almost to the point that I prefer them over the categories as a means to retrieve information. I also use del.icio.us (large scale) for my own bookmarks and to search other people’s when traditional search engines fail. I too fear the day marketers spoil it.

I think you would find that middle size folksonomies work well in niche areas, but fail when pressed for general use. I would also say that folksonomies used on websites trying to cash in on the digg craze fail when they fail to capture a large and diverse user base.

Regarding a “navigation system”, one idea I would love to try out is a two-pane navigation method where you have a tag cloud on one side and the related links on the other. As you click on the a tag, unrelated links are removed from the link list and the cloud is rearrange based on the other tags used on the now refined list of links. The process repeats as you whittle your way down the tag “hierarchy”. Use a bread crumb for the obvious traceability reasons. Thrown in some crafty animations and effects (like links being removed are puffed away) to help users understand what is happening and viola, a dynamic line is drawn from your start to end positions.

BTW, I would like nothing more than to see this idea in the flesh, so to speak. I don’t have time to create this, so consider this a CC share-alike idea.

A single folksonomy presented like interlinked plural taxonomies.

With a Google suggest-like feature?

Put a website taxonomy as the structured view of the world and supplement it with a folksonomy to give us the unstructured view of the world.

Axure, a website prototyping tool, use a folksonomy as the main way of accessing their help files – very useful!

M

Raw Sugar does a better job than many taxonomies I have dealt with.

I have talked with a couple people with tagging system on their intranet that in two months have out paced their taxonomy they the taxonomy it took a year to build.

The real answer is taxonomy and folksonomy need each other.

Just like any other SEO, i agree with you vanderwal, they need each other..Please check out my music/lyrics site and tell me what you think – http://www.mylyricscentral.com

stock.xchng (www.sxc.hu) is a very popular and useful site for stock photography. It contains a mixture of commercial and free (Creative Commons) user-uploaded photos. Users tag with their own keywords when uploading photos.

A good question, Josh. I have caught myself saying that current folksonomy interfaces – not the underlying technology, but rather the way folks are currently viewing and acting on folksonomic structures – are pretty relentlessly text-based. They have some problems, therefore, not unlike command line interfaces.

We probably could use some GUI affordances in folksonomy design. Tag clouds are a teensy first step, allowing scale to reflect some data point like tag frequency. But we could probably do a lot better…

Any examples out there of GUI’d folksonomies?

I believe it is even too easy to see how both folksonomies and taxonomies present important issues (they are features at the same time as well).

Basically taxonomies are not able to represent the varied, multidimensional, ever changing nature of the human way of assigning words to meanings and they are expensive.

Folksonomies on the other side are not able to scale (not only for you but also for other people to navigate the stuff you and others like you submit), not usable, hard to browse and quite useless if you are searching for information following a known-item approach.

That’s why Thomas is completely right, the answer is in the middle and it is a metadata ecology.

I’m reporting below a part of what I have commented on the same topic on David Weinberger blog. Rawsugar is one of the examples but not the only one:

Please, please, please stop saying that folksonomies and taxonomies or that folksonomies and the semantic web are two different, diverging world.

I’m talking and writing about this since the beginning of 2005. Quite since the birth of folksonomies and I’m pretty sad to see that people are still discussing if something like that can exists at all.

It exist. It’s already here! It brings huge benefits! It’s not me saying this, it’s a number of examples already online that give us the proof. A few people call this marriage a middle ground, a metadata ecology (me too).

A few examples:

Etsy is a wonderful example of how taxonomies (a flat topic set) and tags (used as user generated sub topics) can be mixed together. But probably the most interesting part is that Etsy is providing a complete set of different access dimensions to their information through amazing flashy visualization clues. These dimensions are implicitly Ranganathan facets for space, time, material, topic, colors, owners, etc.

RawSugar is a great social bookmarking tool, a sort of del.icio.us on steroids, with a powerful blogging integration tools, ajax and a hierarchical organization of tags. The result is a forest of tag trees than can be used to support wayfinding both in search and navigation.

LibraryThing and PennTags let users catalogue and tag books. Here subject headings are shown side by side with people assigned tags. Again, you can browse using a top down approach or fly horizontally leveraging tags. LibraryThing also has user generated synonyms for tags..

WineLog is a “collaborative wine rating, sharing and tagging site designed to help you keep a record of wines you’ve tried and discover new wines”. Wines are organized and can be browsed through facets (varieties, regions and wineries) and user submitted tags.

And finally we have Facetag.org, the project I presented at the EuroIA in Berlin and soon at the IA Summit in Las Vegas. Probably the fist attempt to integrate bottom-up tagging and top-down traditional classification schemes (hierarchical facets).

Why doing that? Will users ever participate? How could the user interface look like? You can find our work on the site, but the main idea is that the actual flat tagging paradigm is basically useless when the system scales. Our approach can limit the impact of polysemy, homonymy, mistypings, basic level variations, singolars/plurals, while improving serendipity, browsability, usability, scalability, etc. How? Introducing a little bit of hierarchies and facets to tagging while maintaining an usable and easy to learn interface. A sort of easy Flamenco + tags.

And I think Facetag is only one of the 1000 possible tradeoffs between the old and the new world..

Joshua, Thomas.. I would like to have your feedback.

Thanks

I agree with emanuele. A mix is the best solution. Adding personalization upon it will even make it more usefull. Google suggest is great and you can put it on top of a controled taxinomy (as large as folksonomy but with controlled like having only 40 different kind of metal not the 340 available in last.fm)
Emanuele I will follow facetag …and please visit U.[lik] (tagging will be added in two weeks).
It’s a great RSSadded ! Congrats.

Very well and thought out post.
Its good to know all the info you have presented here and I will be looking further into this myself.