During my trip to ICASSP for participating on a panel related to the future directions of Internet, I had discussions with panelists (from Yahoo, Google, and Micrsoft) and also had a chance to discuss with some other researchers abot the efficacy of tags. This was not planned, but just happened. It was clear from this discussion that people are finally finding out and even getting convinced that tags were clearly a fad encouraged by some early successes. People find that tags do help in some cases, but in most cases they only take you so far. This was more in the context of images, video, and audio.
Clearly tags can help a bit, particularly because the current environment is so strongly text-oriented and tags are good example of using text. As discussed earlier on this blog, the idea of text pre-dates current search when technical articles used to ask for keywords to classify them to help librarians and users in (manual) information retrieval. And that never helped much.
Â
My concern with tags is really not with the limited applicability of tags but the distraction that it could, and have, caused. Remember the famous saying that to get to moon it is not much useful to climb the tallest tree (or mountain). Tags will help a bit, but the real problem is representation and utilization of context and using it correctly with visual content. I believe that we have to learn to deal with this. And earlier the focus goes to real problems, sooner we will address the real issues.
By the way, my friend Arun Katiyar, CEO of Seraja, sent me a link from YouTube that is funny and related to the issue of tags. You may want to see that: