What one information junkie wants

I love listening to podcasts when I fly. As I type this, am listening to October’s Semantic Web Gang, where they are talking about Twine. My feelings about Twine are mixed. The concept and, at least every time I hear Nova Spivack talk about it, vision seem consistent with a lot of my interests and needs, but the implementation and execution is mixed to say the least. So I started thinking about what I might really want, and here is what I would want from a service build around my interests.

Let’s assume that I can pipe into Twine any links I share on Delicious and Twitter, the kinds of stories I like and comment upon on Friendfeed, and potentially reach into other sites that I use, especially Freebase which now sits on the linked data web. What I want from this is a good recommendation and discovery engine. I consider Freebase one of the best examples of the power of structured data on the web, especially as a resource. I consider/want Twine to be a discovery service ( I use Twine almost as a placeholder here) that fills a similar niche. Something that combines the power of linked data with the kind of machine learning capabilities that make the Amazon recommendation engine as good as it is, and allows people like Toby to do cool stuff with the data in Freebase. I don’t want it to be a search engine (Lijit does a pretty good job there), but an engine for me to find material I might not find otherwise using Google, Lijit, delicious or Friendfeed. Ideally I want that capability to be exposed as an API which can be leveraged in other, innovative ways (it looks like that might be on the cards)

What I don’t care about is how it works, as long as it is web native and compliant with standards. Of course, RDF etc are technologies that I would like to see succeed and believe in, but this is not a requirement.

As an aside, I wonder if Freebase types can become a generalizable, global ontology which can be exposed to any data source, or would that just not work at scale?