Rethinking information
21 October 2007On the latest blogpost of Think Differently!! they posted a video by Michael Wesch to explain the shift from paper-based information to digital information. It’s a video that tries to summarize some of the content of the book Everything is miscellaneous.
The video sparked some thought about tagging and the way its evolving. Currently tagging is a manual process, users try to identify to what taxonomy the content belongs to. This allows most of the available application to determine what other content is relevant. However most people have different background, cultures, and give different meanings to words. Also people change over time, changing their tagging behavior. So how to overcome this? There seem to be some “automatic” tagging solutions available:
- Jiglu is a tagging solution that indexes page and determines the tags that each page should have, linking the pages together. This solution is especially convenient for all those large blog’s out there.
- Twine “applies a semantic analysis to it that creates tags for each document, video or photo. The tags match up to concepts that Twine’s algorithms associate with each piece of content, regardless of whether that concept is specifically mentioned in the web page or other content being tagged. What Twine basically does is automatically generating smart tags and connects them together.” as explained by ERP-network.
The next step will be to personalize the tags based on the people’s browsing behavior. Computers have become stronger, grid computing isn’t that expensive anymore, so the obvious step becomes to integrate machine learning into web applications and use it for personalization. Still, I have a feeling that somehow we are not there yet. Its still difficult to make tags meaningful, linking all kinds of resources together doesn’t make them meaningful but makes them more useful. Meaningful would be if we could integrate a process of negotiation with the network/community into the tagging. If tagged content could evolve over time based on newly tagged content. And if tagged content would respect the amount of details the people requires (a person with a lot of information about programming might like to specify it further while a manager could be satisfied with just a programming tag).
28 October 2007 at 21:26
Hi Abdul,
here is what I would like to see, a system that when I add a tag, points me to the tags that share a maximum of similar content to mine, and has a capabality to interconnect them, so that we can get ecologies of similar tags; This would allow us to create an alerting system, so that new items, in order of the largest number of choices for example, would be listed …
Is that on the horizon anywhere?
Michel
28 October 2007 at 21:34
I didn’t even think of that, content associated with the tag is what defines the meaning of the tag(s). But current systems (as far as I know) are not there yet, they extract the similarities between documents, and add tags that they think should be associated with that. So similar documents get linked in that way.
Thnx, I really need to think this through both conceptually as technical! Only question still remains (forme), how would you notice if the meaning of the tags shifts, till when do you remember the old meaning (given by the content) to the tag?