I wonder if you could generate this automatically. Let me explain where I'm coming from. I made the Theory of Computing Blog Aggregator (http://feedworld.net/toc), and part of the reason I didn't set up more aggregators with the software was that you need domain-knowledge to create the blogrolls.
My idea was that you could start from 2 or 3 "seeds" on a given topic, find which blogs they are linking to, filter them by pagerank or technorati rank or whatever, and use an off-the-shelf machine learning tool to determine if they are on the same topic. I never got around to it; other projects I'm working on took up all of my time.
But if you can build this intelligence, you can create aggregators/memetrackers/whatever for an unlimited number of topics. And further create communities around each one and so on.
If someone wants to collaborate on this idea, let me know.
P.S. To build my aggregator, I started off with planet planet (http://www.planetplanet.org/), but made a bunch of improvements like automatic comment import. There is an unreleased version with a lot more features (including a google reader-like UI) that I will be happy to show anyone who's interested.
Incidentally, one of the data sets that we have from a friend of ours that we're playing with internally is a collection of links from the entire German blogosphere. From there we can do queries to see which blogs are similar to others, so if you had a starting point of blogs that you liked, it can suggest others.
Completely unrelated, a while back we did start putting together a planet for some of the folks from HN. It didn't really take off, but there's still some good reading there:
In a perfect world, that's exactly how I'd do it, too. Right now though, I'm still working out the bugs on how to get the mix of sources right and make the presentation interesting. Machine learning is being applied to which sources make the front page and lead the memes, based on clicks from readers. Ideally, the "picking up" of sources would also be automated as well.
My idea was that you could start from 2 or 3 "seeds" on a given topic, find which blogs they are linking to, filter them by pagerank or technorati rank or whatever, and use an off-the-shelf machine learning tool to determine if they are on the same topic. I never got around to it; other projects I'm working on took up all of my time.
But if you can build this intelligence, you can create aggregators/memetrackers/whatever for an unlimited number of topics. And further create communities around each one and so on.
If someone wants to collaborate on this idea, let me know.
P.S. To build my aggregator, I started off with planet planet (http://www.planetplanet.org/), but made a bunch of improvements like automatic comment import. There is an unreleased version with a lot more features (including a google reader-like UI) that I will be happy to show anyone who's interested.