what is the next step for tagging?

The Web needs tagging or some other bottom up and scalable system for managing the tons of information that everyone is creating. Librarians and specialists will not be able to keep up with classifying the tsunami of information, and we’ll all lose out if we don’t find some way of capturing the patterns of behavior that people sharing information create: there are business needs for sure, but mostly there is a lot of general improvement to people’s lives that can be had (e.g. if I subscribe to popular links for a tag on del.icio.us, I know most everything that people are thinking about a topic on a real time basis, and I am included and benefit from thousands of small intentions and thoughts).
So the goal is a good goal. Tagging has become sort of popular, but some problems have started to crop up. Many people still are unable to see the value of the practice, and for those that do tag there are limitations built into the standard model (it works much better as a tool for collaborating than for personal organization, and you spend a lot of time reinventing the classification wheel). In response, some geeks have started talking about extending the model to include relationships between terms or normalization, or ranking through collaborative filtering. That’s natural (in fact, it has a precedent; in the 50s there was a single-term, non- hierarchical classification system for libraries called UniTerm, but it was not generally accepted and instead we have the hierarchical Library of Congress and Dewey Decimal systems). Time Tags and Zone Tags are interesting ways of adding layers of social meaning to tags.
This direction feels wrong to me though. Tagging has had success because it is lightweight enough metadata to be easy, and it’s non-hierarchical nature lends itself to social applications (like quick aggregation of photo themes on flickr). I think a better approach will be to build tags into the interaction model of applications, turn the actions and intentions of users into inferred or implied tags, then surface that information as a basis for explicit meta-tagging action later (instead of putting so much of a burden on the user).
Right now, for example, a searcher types in one or two words as a query to a search engine, then they might pick one or two Web pages to visit. Essentially, they have tagged those pages with the query term, but that metadata is currently lost to the searcher. Instead we ask the searcher to take explicit action to save a Web page, then tag it with terms that may have nothing to do with their original query; the relationship between the two metadata linked to the same entity could be very valuable to both the user and the system. I think this kind of explicit/implicit tagging could add a lot of relevance and richness to a lot of sites and applications.

Post a Comment

Your email is never shared. Required fields are marked *