Folksonomies shape search behavior

With the proliferation of tagging based services such a del.icio.us and Flickr, the term folksonomy has started to pop up in discussions about Web content. It's a fairly new term that has grown organically out of the need to distinguish between the formal classification of information done by experts such as librarians and the more casual categorization that users and authors of Web content do when they describe content with keywords.

Folksonomy is a derivative of the term taxonomy. Information architects use taxonomies to classify information according to a predefined formal scheme. In a taxonomy there is typically a hierarchy between the items it attempts to describe. For example, fuel is a parent concept for diesel oil, gasoline and butane. Folksonomy, on the other hand, has no explicit hierarchy. It is a flat namespace without parent-child or sibling relationships.

Flickr is an example of a folksonomy based system. When amateur shutterbugs take photos and upload them to Flickr, they are asked to describe each shot with keywords. The keywords are completely of their own choosing. No existing categorization scheme is given or even suggested. One photographer might tag her shot of Paris with a keyword "eiffeltower", while another might tag his with "girl", "ladefense" or "arch".

Similarly, Technorati does not provide bloggers with a taxonomy by which to tag interesting Web sites. What the two systems do provide, however, is clusters of related tags. The tag "paris" has related tags "France", "Hilton" and "travel" in Technorati, derived from posts submitted by other users.

The concept of related information has a crucial impact on users' search behavior. Information on the Web is uncategorized. It is hard even for Googlebot to figure out the interrelationships between documents; it couldn't possibly index all the new content according to any formal taxonomy. There is simply too much content and it is too varied. Folksonomy based sites provide a creative solution to the problem: they encourage the user to browse related information to find interesting content. By following the links of related tags a Technorati user will likely find blog posts he is interested in but which he would not have thought of searching for with a specific Google query of his own devise.