Academic Search Engine Published
from the I-Still-Haven't-Found-What-I'm-Looking-For dept
Google determines relevance by links, and so does this search engine in Nature. However while Google only counts a link from site A to site B as a single vote for site B, the published algorithm somehow takes into account all of the links on site A to establish a “community” of links. I think Google could use some improvements, just search for “Russian Dancing” sometime. Why are my roommates the first hit?
Comments on “Academic Search Engine Published”
seems 'not new'
If there’s something new in this work, it seems like a minor difference from a) kleinberg’s HITS algorithmn; b) NEC’s citeseer algorithm; c) other works explicitly on finding communities by looking solely at the link structure (Prabhakar Raghavan, for one).
Maybe ‘what’s new’ is that it’s published in Nature, but most of the important ideas (on which this is based) were published years ago, as far as I can tell. (At least the article could mention some more of the prior work).
Here’s the researcher’s homepage:
Hmm. Seems as good a place as any to link to the story that all the blogs have been talking about lately to make themselves feel more important about how blogs impact Google. I know there are plenty of random Google searches that lead people to Techdirt for no clear reason – but it’s fun to read the search logs.
Mike, you forgot to mention the “truckload of vegetables” argument which seems to come up here a lot recently. And on Google, too.