Phil Butler wrote an interesting article on R/WW.
http://www.readwriteweb.com/archives/powerset_and_hakia_quest_for_semantic_web.php

I wrote this comment on it...

Not so sure freebase is the semantic web. They are really providing a mechanism for structured data to be stored consistently. The challenge of a semantic search system is to structure the currently unstructured mess that is most web pages - the content people want to find. Freebase doesn't do semantical language analysis like hakia and powerset are trying to do.

The semantic web is really a reflection of the structuring of loose data. That'll come from two directions: web sites wanting to do it (formatting their information with microformats is the most likely course I think) and semantical conversion engines, like these two, trying to brute force that conversion. Ultimately you need both, but it's going to take a while to see real results.

Which brings up the thought: is there a product out there yet that provides a driving value to web sites to semanticise (is that a word?) their data. If the web sites spend the time doing this themselves it would go a long way to making progress towards more structured information on the web.

Sounds like an idea for a startup in there somewhere...

Unfortunately no, and it's largely because people don't understand the potential of the semantic web. Yet.

And a big challenge with semantic web apps is ontologies. You are dismissing the benefit of storing data consistently that freebase does - the benefit of a semantic web is so that you can find data consistencies and this is very valuable. Having said that though, I can't imagine freebase working out. I don't see how a person would be motivated to "structure" the data - they seem to think that if people are willing to edit wikipedia, then they must be motivated to structure sentences into types. Adding content to wikipedia is a different experience than structuring it, and I think these guys don't fully understand the motivation for people to edit wikipedia. The fact threre is a semantic mediawiki, pretty much makes freebase irrelevant.

Radar Networks will be interesting to watch when they (apparently) will launch later this year. They are essentially building a consolidated wiki, blog and e-mail application that a user can store all their data in it. It could work because people will have see a benefit in semanticising their data.