Friday, March 07, 2008

Social Semantic Tagweb Ontology is not a Dewy Decimal System Taxonomy

I stumbled upon a very interesting presentation that Tom Gruber gave at The 5th International Semantic Web Conference about how the Social Web and the Semantic Web are not two opposed visions for the web, moreover, they are so compatible that we can form a Semantic Web that enhances our Social Web experience, and the Semantic Web can be more meaningful with the input (the Collective Intelligence) of people on the Social Web.
He raises many interesting points such as:

  • The difference between taxonomy, ontology, and folxonomy and how they relate. He especially points out that the goal of the Semantic Web isn't to create a taxonomy (a rigid categorization of the web), but an ontology (a means for different data from different sources to be understood, related, and used).

  • He observes that tagging consists of a relationship between an entity (tagget), a URI (what is being tagged), and a phrase (the tag).

  • There are issues that arise in tagging (having to do with each of those entities), especially when one wants to search through tags from different websites:

    • Tag spelling, whitespace, capitalization, and pluralization - which "different" tags should actually be considered the same?

    • What if there are different URI's for the same "thing": How do we detect this? What do we do about it?

    • Different kinds of taggers: What if we let machines tag things - should those be considered to tags by humans?

    • What happens when you change your mind: what if something is un-tagged?

    • Tag polarity: Should users be able to tag things as not-something?


  • Another interesting concept presented is the idea that browsing can be viewed as browsing along the dimensions of a hypercube: location dimensions, tag dimensions, time dimensions, etc.

  • Useful things can be done like letting tags be inherited: if a picture was taken in Paris, it was also taken in France - so we can add the tag "France" to anything tagged "Paris", right? (what if it's Paris, Texas... or some other Paris?).

  • More fun topics like structured data and how with Google we are getting locked into the string-of-words style of searching.

In short, there is a lot of focus on the potential of improving tagging systems, and on the potential of the future of tagging and the web in general.

If you have an extra hour of nothing to do, or if you just want to put it on in the background while you're doing chores, it's definitely worth watching/listening to.
I'd also be glad to hear your thoughts on this.

1 comment:

Median Joe said...

Thanks Yuriy, I will definitely set aside time to watch the presentation. I also believe there is a lot of untapped potential, from visualizing and interacting with tags assertions, to inference (especially with semantic hierarchies of tags), and also with improving the search experience.