Random blog-like rambling from Rachel's brain. A mixed up mess of usability posts, fiction, and travel.

On Folksonomies and Taxonomies

Any site that collects large amounts of information benefits from having some classification means that helps users find what they are looking for. Take the library, for example, if you are seeking a book on say, zoology, look no further than the card catalog and you'll find all the books you need within the carefully constructed taxonomy of the Dewey Decimal System. Categorization keeps us from blindly hunting through haystacks in search of needles.

We've gotten used to the Dewey Decimal type systems of the world. Traditionally, even on the internet, the best way we could think of to categorize information was to put the power in the hands of skilled information architects to create taxonomies by which we could organize our piles of information. Now though, enter Web 2.0 and the tag.

Firstly, some definitions.

Taxonomy: a particular classification arranged in a hierarchical structure.

Folksonomy: a system of classification derived from the practice and method of collaboratively creating and managing tags to annotate and categorize content.

Taxonomies are old and reliable methods for organizing content, folksonomies on the other hand are a new breed that came about with the advent of social bookmarking sites like delicious and photo management sites like flickr.

Both methods have their strengths and weaknesses, so now, when you're creating a new site that requires robust organization you have a decision to make: taxonomy or folksonomy or both?

The Taxonomic Approach

Taxonomies are managed. That means they have a strict structure and are completely predictable. When well designed they fit all of the content required of them and make for an easy navigational structure.

Netflix's Taxonomic Approach

Netflix's Taxonomic Approach

Sites like Netflix, Amazon and eBay generally benefit a great deal from using a taxonomic approach for organizing the content. A clean hierarchical structure makes for a clean navigation.

One of the reasons these sites work is that the information being organized tends to fit very well into defined categories and furthermore, those categories make immediate sense to the users of the sites.

That said, there are always places where these schemes fall short. It is up to a small group of people to define these categories, and from time to time where an item fits in the scheme is not terribly apparent. Sometimes, an item appears to fit in more than one place making it harder to find. The overlap between, say "Crime Thriller" and "Drama" is not trivial.

You can overcome that by placing items in multiple categories, but there's always an issue here with anticipating how users are going to see things. Putting together a proper taxonomy is time intensive process that ideally utilizes interviews with the user community. Not everyone has the time or the money to do taxonomies correctly. A bad taxonomy is almost worse than no organization at all. Think of it as searching for needles in 500 haystacks all claiming to have needles but that actually just have hay.

Folksonomies: A Saving Grace?

In an effort to avoid the pitfalls of a bad taxonomy, it's easy to think that folksonomies are the way to go. In this approach, you leave it on the shoulders of your user community to gradually build it's own unique organizational scheme. As content is added or discovered, the users are meant to tag that content with meaningful keywords that will help other users find that content.

The approach works best when users are adding the content themselves. On Flickr, I upload an image and quickly add some defining tags that will help the community locate it, and similar files.

Flickr's folksonomic approach

Flickr's folksonomic approach

Tags can be extremely powerful, but they don't necessarily solve all your organization problems. Relying on a user community to drive organization leads to some pitfalls. For example, there is no team in charge of managing the scheme and so there's high likelihood that you'll see misspellings, variations and overlapping tags that can make finding just the right piece of content quite difficult. Say I'm looking for cat pictures, but the perfect image hasn't been tagged "cat" and was instead tagged "kat". The possibility of my locating the image is small without expending a great deal of effort in my hunt.

Even if you could guarantee that every tag was spelled correctly it won't take long for a folksonomy to grow so large that it becomes unwieldy. For awhile when this concept was first introduced you often saw on sites employing it the Tag Cloud. Clouds were interesting, showing you the popular tags using font size to bring the popular ones to the fore front. This worked fine for some sites, but once there are so many tags clouds become relatively meaningless. Given the sporadic ways that users tag items it isn't uncommon to see a folksonomy that has 1000 tags, each corresponding to a single piece of content. Such an organizational structure does not really assist a user in locating anything. It would be just as easy to search a list of each item as it would be to scan the tags.

A Combined Approach

Ideally, it may be that the way to navigate the issues inherent in either approach is to push taxonomies and folksonomies together. Your taxonomy might not be perfect, but if you allow users to add tags to supplement that taxonomy it gives the community another route to the information. In addition, such an approach actually can work as a means to gather important feedback on your taxonomy itself. Those tags users add in to help you out are bits of invaluable feedback that can be used to fine tune the schema.

Regardless, each website has particular needs that have to be carefully examined before any one approach is decided on. As usual there isn't really a strict right or wrong way to do things. Everything relies on context.

Now, I might just need to go and create my own taxonomy for future Usability Shark postings. I've done a little internal debating of my own on whether it could be helpful or not. I'm leaning now, towards yes.