New tech ranks pages without links

Are they talking about Blei's LDA algorithm? AFAIK it is the most influential algo in recent times (2700 citations in 8 years!). http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.80....

kind of like an LDA portal:

http://www.arbylon.net/projects/

I had no idea what the news article was talking about, but it sounds like topic modeling with a temporal component. Why didn't they put the paper title or algorithm name in the news article? It's frustrating to read something dumbed down enough that you can't even recognize it.

Update: I am betting it is this paper - "A Language-based Approach to Measuring Scholarly Impact" http://www.cs.princeton.edu/~blei/papers/GerrishBlei2010.pdf

This would only work for a distinct set of documents/websites, correct?

If an article completely duplicated the content of another article, wouldn't this algorithm treat the same almost identically, if it does not use "links" of some kind?

While this is really promising - it's important to note that they experimented on this using only journal articles. It'd be interesting to see how it would deal with spam, as it seems like this is the sort of thing that could be gamed.