Geospatial Semantic Web Blog

Today’s conventional tagging mechanism, found on flickr, del.icio.us and youtube, allows users to associate web resources (i.e., documents, photos, videos and audios) with a set of keywords (i.e., tags). This mechanism works well for tagging resources that have a direct accessible URL. However, it falls short in being an effective tool for labeling the “deep” content of those resources — e.g., a partial clip from a youtube video.

To solve this problem, startups are building around a new tagging idea called deep tagging. Deep tagging is extremely useful in organizing and sharing lengthy video and audio files online. Here is how deep tagging works in Veotag:

1. Users upload a video or audio file to Veotag
2. Using a web-based deep tagging tool, users can assign tags (veotags) to specific clips from the file.
3. When other users are viewing this file, they have the option to play the whole file from the beginning to the end or jump back and forth between different veotagged clips.

See this Veotag video for more information.

I’m intrigued by this new deep tagging idea.

First, implementations of this idea prove to be successful in solving immediate information sharing problems on the Web. Take Veotag as an example. Let’s say that you are watching an hour long keynote speech. With veotags, you will be able to jump to the specific part of talk that interests you.

Second, the idea of deep tagging seems to converge with the idea of the Semantic Web. Both ideas share a common goal: to give explicit meanings to hidden information on the Web, so that knowledge can be better organized and shared. Veotag and other similar services have successfully demonstrated the value of user-generated metadata, and showed why an everyday user may want to spend time and effort to markup resources.

Looking forward into the future. I speculate that we will see more Web 2.0 sites to come with deep tagging support. It’s an inevitable part of tagging evolution — people begin with labeling things as a whole, and latter they learn to label parts of a thing. The next step in deep tagging is to allow users to make assertions about video and audio clips using expressive markup languages such as RDF and OWL. While today’s deep tagging (e.g, veotags) is useful to humans (e.g., you and me), tomorrow’s semantic deep tagging (e.g., triples) will be useful to robots (e.g., Google bots).