Last week I mentioned a conversation with David Lankes that touched upon folksonomies and the idea of tag decay. Folksonomies are a subject near and dear to my brain, and Dr. Lankes made me aware of a completely new dimension to them.
When I talk up the concept of user-created tags, I often tell folks that it’s a way for those within the libraries to get an idea of how the users are thinking. After all, when I’m on the ref desk showing a patron how to find items on a particular subject in a catalog or database, I’ll usually tell them to do keyword searches to track down a single item that meets their criteria, and then look at the subject headers. These headers will indicate how the database creators categorized the item, and those categories can then be used to quickly find other similar items.
So, it should work the same way in reverse, right? If librarians look at how users tag an item, it can give us a parallel insight into how users think!
Except, as Dr. Lankes pointed out, a user that places a tag on an item may not consider that tag appropriate a year later, as that user’s knowledge, experience, and relationship with the item change. So the picture tags give us gets blurred.
Enter the concept of tag decay*. Tags are no longer permanent, but fade over time; this could mean that its relevance slowly decreases to nothing, or maybe every tag gets a simple expiration date. Such a practice would raise other issues, such as users assuming their tags are permanent. For example, if I still think a resource should be marked “NYS government,” how will the system let me know that the tag I placed in 2007 is no longer valid? Is such a system even feasible without some kind of user authentication, and would such authentication prove too great a barrier to use?
User-created tagging is a technique that I think holds great potential for libraries who want to know more about their users (this is why I mention the Powerhouse Museum so frequently); even without something like tag decay, tagging is a powerful tool for users to customize their experience with resources and to communicate with those who maintain those resources. Addressing the disconnect between permanent tags and impermanent preferences would only help to better meet that potential.
* I found a researcher — Terrell Russell — who uses this term to refer to the growing staleness of tags over time. To clarify, in this post (and in the conversation it describes), the term is used instead to refer to a system that would allow older, obsolete tags to fade into irrelevance automatically. If I write on this again, I may follow Joshua Porter‘s lead and refer to the the other phenomenon as “popularity decay”, for clarity.