Tags for the masses, ontologies for developers
→ Home

In my line of research I?m very much involved with ontology development. I?m not going to beat around the bush: developing ontologies is hard. Really hard. The more logically rigorous they get, the more difficult they become to construct.

So, you might ask, how is the vision of the great and wonderful ?semantic web? ever going to work? After all, ontologies are the framework that is meant to undergrid the Internet of tomorrow.

Take a look at del.icio.us, flickr.com and technorati. They all use an up-and-coming (craze of the moment) idea of tagging. You allow people to add any word to their content and collect all these tags up into a large list. The larger the font, the more frequently used the tag. The obvious problems are synoyms and homonyms. However: who cares?! It kind of works, anyone can understand the idea, so wa-hey: let?s go tag crazy.

Ontologies however are much more powerful and dangerous. They exactly and unambiously define terms and formally capture relationships between terms. You get transitivity, inheritance and other great stuff like that. Moreover, computers can automatically navigate these data structures and use them to answer almost any question you can you throw at them. Feel the power!

What to do? The general populus is never going to be able to author ontologies, but could possibility be induced to use them, given a simple enough interface. So, if the subject area we are describing is sufficiently limited that we can construct an ontology to cover it (no one is going to be able to create an ontology of ?everything??), then we can allow people to tag their content with our ontologies terms. The result: we can have our computers sort, manage, slice and dice their tagged content any which way, take advantage of all the advanced features and the world is a better place. Amen.