Talked with an engineer/coworker about the concept and some analogies to our interacting multiple model tracker. He mentioned hitting multiple semantic APIs at the same time and merging their information to get a better answer. loved the concept.
He was a psych graduate before going back and getting a physics degree so he also brought up sarcasm and keying in on the nuances of characters like: 
, bolds, italics
I mentioned we will have to rely somewhat (and leverage off of) the social media conversations where interactions occur (again stuff to think about after the prototype but highly relevant)
Vlad 4:42 pm on June 29, 2009 Permalink
That’s good stuff. I think it will be attention to details like that which will really set our solution apart and create real value for both consumers and advertisers.
I had some far out thoughts at well today – this is probably way down the road but if we could integrate visual contextual recognition from users’ Facebook or Flickr photos, then we have an additional layer of information. For instance if someone uploads an image of themselves at a party drinking out of a Budweiser bottle that gives advertisers and our engine a huge amount of information about the consumption habits of that user. It’s just a thought